train scripts history

This reverts commit 648cf32e9c.

.gitignore

@@ -10,3 +10,13 @@ poetry.toml
 **/log/
 *.spec
 dist/
+*steps.zip
+*.pkl
+best_model.zip
+*.csv
+*.npz
+*.xml
+*.json
+*.yaml
+*.iml
+*.TXT

command.md

@@ -0,0 +1,14 @@
+训练（默认）
+bash train.sh
+
+测试（实时+显示画面)
+GYM_CPU_MODE=test GYM_CPU_TEST_MODEL=scripts/gyms/logs/Walk_R0_005/best_model.zip GYM_CPU_TEST_FOLDER=scripts/gyms/logs/Walk_R0_005/ GYM_CPU_TEST_NO_RENDER=0 GYM_CPU_TEST_NO_REALTIME=0 bash train.sh
+
+测试（无画面、非实时）
+GYM_CPU_MODE=test GYM_CPU_TEST_NO_RENDER=1 GYM_CPU_TEST_NO_REALTIME=1 bash train.sh
+
+retrain（继续训练）
+GYM_CPU_MODE=train GYM_CPU_TRAIN_MODEL=scripts/gyms/logs/Walk_R0_005/best_model.zip bash train.sh
+
+retrain+改训练超参
+GYM_CPU_MODE=train GYM_CPU_TRAIN_MODEL=scripts/gyms/logs/Walk_R0_004/best_model.zip GYM_CPU_TRAIN_LR=2e-4 GYM_CPU_TRAIN_CLIP_RANGE=0.13 GYM_CPU_TRAIN_BATCH_SIZE=256 YM_CPU_TRAIN_GAMMA=0.95 GYM_CPU_TRAIN_ENT_COEF=0.05 GYM_CPU_TRAIN_EPOCHS=8 bash train.sh

communication/server.py

@@ -1,5 +1,6 @@
 import logging
 import socket
+import time
 from select import select
 from communication.world_parser import WorldParser
 
@@ -10,15 +11,27 @@ class Server:
     def __init__(self, host: str, port: int, world_parser: WorldParser):
         self.world_parser: WorldParser = world_parser
         self.__host: str = host
-        self.__port: str = port
-        self.__socket: socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
-        self.__socket.setsockopt(socket.IPPROTO_TCP, socket.TCP_NODELAY, 1)
+        self.__port: int = port
+        self.__socket: socket.socket = self._create_socket()
         self.__send_buff = []
         self.__rcv_buffer_size = 1024
         self.__rcv_buffer = bytearray(self.__rcv_buffer_size)
 
+    def _create_socket(self) -> socket.socket:
+        sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
+        sock.setsockopt(socket.IPPROTO_TCP, socket.TCP_NODELAY, 1)
+        return sock
+
     def connect(self) -> None:
         logger.info("Connecting to server at %s:%d...", self.__host, self.__port)
+
+        # Always reconnect with a fresh socket object.
+        try:
+            self.__socket.close()
+        except OSError:
+            pass
+        self.__socket = self._create_socket()
+
         while True:
             try:
                 self.__socket.connect((self.__host, self.__port))
@@ -27,12 +40,19 @@ class Server:
                 logger.error(
                     "Connection refused. Make sure the server is running and listening on {self.__host}:{self.__port}."
                 )
+                time.sleep(0.05)
 
         logger.info(f"Server connection established to {self.__host}:{self.__port}.")
 
     def shutdown(self) -> None:
-        self.__socket.close()
-        self.__socket.shutdown(socket.SHUT_RDWR)
+        try:
+            self.__socket.shutdown(socket.SHUT_RDWR)
+        except OSError:
+            pass
+        try:
+            self.__socket.close()
+        except OSError:
+            pass
 
     def send_immediate(self, msg: str) -> None:
         """
@@ -72,37 +92,37 @@ class Server:
         self.commit(msg)
         self.send()
 
-    def receive(self) -> None:
-        """
-        Receive the next message from the TCP/IP socket and updates world
-        """
+    def receive(self):
 
-        # Receive message length information
-        if (
-            self.__socket.recv_into(
-                self.__rcv_buffer, nbytes=4, flags=socket.MSG_WAITALL
+        while True:
+
+            if (
+                    self.__socket.recv_into(
+                        self.__rcv_buffer, nbytes=4, flags=socket.MSG_WAITALL
+                    ) != 4
+            ):
+                raise ConnectionResetError
+
+            msg_size = int.from_bytes(self.__rcv_buffer[:4], byteorder="big", signed=False)
+
+            if msg_size > self.__rcv_buffer_size:
+                self.__rcv_buffer_size = msg_size
+                self.__rcv_buffer = bytearray(self.__rcv_buffer_size)
+
+            if (
+                    self.__socket.recv_into(
+                        self.__rcv_buffer, nbytes=msg_size, flags=socket.MSG_WAITALL
+                    ) != msg_size
+            ):
+                raise ConnectionResetError
+
+            self.world_parser.parse(
+                message=self.__rcv_buffer[:msg_size].decode()
             )
-            != 4
-        ):
-            raise ConnectionResetError
 
-        msg_size = int.from_bytes(self.__rcv_buffer[:4], byteorder="big", signed=False)
-
-        # Ensure receive buffer is large enough to hold the message
-        if msg_size > self.__rcv_buffer_size:
-            self.__rcv_buffer_size = msg_size
-            self.__rcv_buffer = bytearray(self.__rcv_buffer_size)
-
-        # Receive message with the specified length
-        if (
-            self.__socket.recv_into(
-                self.__rcv_buffer, nbytes=msg_size, flags=socket.MSG_WAITALL
-            )
-            != msg_size
-        ):
-            raise ConnectionResetError
-
-        self.world_parser.parse(message=self.__rcv_buffer[:msg_size].decode())
+            # 如果socket没有更多数据就退出
+            if len(select([self.__socket], [], [], 0.0)[0]) == 0:
+                break
 
     def commit_beam(self, pos2d: list, rotation: float) -> None:
         assert len(pos2d) == 2

scripts/commons/Server.py

@@ -1,15 +1,16 @@
 import subprocess
 import os
+import time
 
 
 class Server():
-    def __init__(self, first_server_p, first_monitor_p, n_servers) -> None:
+    def __init__(self, first_server_p, first_monitor_p, n_servers, no_render=True, no_realtime=True) -> None:
         try:
             import psutil
             self.check_running_servers(psutil, first_server_p, first_monitor_p, n_servers)
         except ModuleNotFoundError:
             print("Info: Cannot check if the server is already running, because the psutil module was not found")
-            
+
         self.first_server_p = first_server_p
         self.n_servers = n_servers
         self.rcss_processes = []
@@ -17,30 +18,50 @@ class Server():
 
         # makes it easier to kill test servers without affecting train servers
         cmd = "rcssservermj"
+        render_arg = "--no-render" if no_render else ""
+        realtime_arg = "--no-realtime" if no_realtime else ""
         for i in range(n_servers):
-            self.rcss_processes.append(
-                subprocess.Popen((f"{cmd} --aport {first_server_p+i} --mport {first_monitor_p+i}").split(),
-                stdout=subprocess.DEVNULL, stderr=subprocess.STDOUT, start_new_session=True)
+            port = first_server_p + i
+            mport = first_monitor_p + i
+
+            server_cmd = f"{cmd} -c {port} -m {mport} {render_arg} {realtime_arg}".strip()
+
+            proc = subprocess.Popen(
+                server_cmd.split(),
+                stdout=subprocess.DEVNULL,
+                stderr=subprocess.STDOUT,
+                start_new_session=True
             )
 
+            # Avoid startup storm when launching many servers at once.
+            time.sleep(0.03)
+
+            rc = proc.poll()
+            if rc is not None:
+                raise RuntimeError(
+                    f"rcssservermj exited early (code={rc}) on server port {port}, monitor port {mport}"
+                )
+
+            self.rcss_processes.append(proc)
+
     def check_running_servers(self, psutil, first_server_p, first_monitor_p, n_servers):
         ''' Check if any server is running on chosen ports '''
         found = False
         p_list = [p for p in psutil.process_iter() if p.cmdline() and "rcssservermj" in " ".join(p.cmdline())]
-        range1 = (first_server_p, first_server_p  + n_servers)
-        range2 = (first_monitor_p,first_monitor_p + n_servers)
+        range1 = (first_server_p, first_server_p + n_servers)
+        range2 = (first_monitor_p, first_monitor_p + n_servers)
         bad_processes = []
 
-        for p in p_list:  
+        for p in p_list:
             # currently ignoring remaining default port when only one of the ports is specified (uncommon scenario)
             ports = [int(arg) for arg in p.cmdline()[1:] if arg.isdigit()]
             if len(ports) == 0:
-                ports = [60000,60100] # default server ports (changing this is unlikely)
+                ports = [60000, 60100]  # default server ports (changing this is unlikely)
 
             conflicts = [str(port) for port in ports if (
-                (range1[0] <= port < range1[1]) or (range2[0] <= port < range2[1]) )]
+                    (range1[0] <= port < range1[1]) or (range2[0] <= port < range2[1]))]
 
-            if len(conflicts)>0:
+            if len(conflicts) > 0:
                 if not found:
                     print("\nThere are already servers running on the same port(s)!")
                     found = True
@@ -55,9 +76,8 @@ class Server():
                     for p in bad_processes:
                         p.kill()
                     return
-            
 
     def kill(self):
         for p in self.rcss_processes:
             p.kill()
-        print(f"Killed {self.n_servers} rcssservermj processes starting at {self.first_server_p}")
+        print(f"Killed {self.n_servers} rcssservermj processes starting at {self.first_server_p}")

scripts/commons/Train_Base.py

@@ -31,31 +31,29 @@ class Train_Base():
         args = script.args
         self.script = script
         self.ip = args.i
-        self.server_p = args.p              # (initial) server port
-        self.monitor_p = args.m + 100           # monitor port when testing
-        self.monitor_p_1000 = args.m + 1100 # initial monitor port when training
+        self.server_p = args.p  # (initial) server port
+        self.monitor_p = args.m + 100  # monitor port when testing
+        self.monitor_p_1000 = args.m + 1100  # initial monitor port when training
         self.robot_type = args.r
         self.team = args.t
         self.uniform = args.u
         self.cf_last_time = 0
         self.cf_delay = 0
         # self.cf_target_period = World.STEPTIME # target simulation speed while testing (default: real-time)
-    
-    
 
     @staticmethod
     def prompt_user_for_model(self):
 
-        gyms_logs_path = "./mujococodebase/scripts/gyms/logs/"
+        gyms_logs_path = "./scripts/gyms/logs/"
         folders = [f for f in listdir(gyms_logs_path) if isdir(join(gyms_logs_path, f))]
-        folders.sort(key=lambda f: os.path.getmtime(join(gyms_logs_path, f)), reverse=True) # sort by modification date
+        folders.sort(key=lambda f: os.path.getmtime(join(gyms_logs_path, f)), reverse=True)  # sort by modification date
 
         while True:
             try:
-                folder_name = UI.print_list(folders,prompt="Choose folder (ctrl+c to return): ")[1]
+                folder_name = UI.print_list(folders, prompt="Choose folder (ctrl+c to return): ")[1]
             except KeyboardInterrupt:
                 print()
-                return None # ctrl+c
+                return None  # ctrl+c
 
             folder_dir = os.path.join(gyms_logs_path, folder_name)
             models = [m[:-4] for m in listdir(folder_dir) if isfile(join(folder_dir, m)) and m.endswith(".zip")]
@@ -64,16 +62,17 @@ class Train_Base():
                 print("The chosen folder does not contain any .zip file!")
                 continue
 
-            models.sort(key=lambda m: os.path.getmtime(join(folder_dir, m+".zip")), reverse=True) # sort by modification date
-            
+            models.sort(key=lambda m: os.path.getmtime(join(folder_dir, m + ".zip")),
+                        reverse=True)  # sort by modification date
+
             try:
-                model_name = UI.print_list(models,prompt="Choose model (ctrl+c to return): ")[1]
+                model_name = UI.print_list(models, prompt="Choose model (ctrl+c to return): ")[1]
                 break
             except KeyboardInterrupt:
                 print()
 
-        return {"folder_dir":folder_dir, "folder_name":folder_name, "model_file":os.path.join(folder_dir, model_name+".zip")}
-
+        return {"folder_dir": folder_dir, "folder_name": folder_name,
+                "model_file": os.path.join(folder_dir, model_name + ".zip")}
 
     # def control_fps(self, read_input = False):
     #     ''' Add delay to control simulation speed '''
@@ -87,7 +86,7 @@ class Train_Base():
     #             if speed == '0':
     #                 inp = input("Paused. Set new speed or '' to use previous speed:")
     #                 if inp != '':
-    #                     speed = inp   
+    #                     speed = inp
 
     #             try:
     #                 speed = int(speed)
@@ -95,7 +94,7 @@ class Train_Base():
     #                 self.cf_target_period = World.STEPTIME * 100 / speed
     #                 print(f"Changed simulation speed to {speed}%")
     #             except:
-    #                 print("""Train_Base.py: 
+    #                 print("""Train_Base.py:
     # Error: To control the simulation speed, enter a non-negative integer.
     # To disable this control module, use test_model(..., enable_FPS_control=False) in your gyms environment.""")
 
@@ -108,15 +107,15 @@ class Train_Base():
     #     else:
     #         self.cf_delay = 0
 
-
-    def test_model(self, model:BaseAlgorithm, env, log_path:str=None, model_path:str=None, max_episodes=0, enable_FPS_control=True, verbose=1):
+    def test_model(self, model: BaseAlgorithm, env, log_path: str = None, model_path: str = None, max_episodes=0,
+                   enable_FPS_control=True, verbose=1):
         '''
         Test model and log results
 
         Parameters
         ----------
         model : BaseAlgorithm
-            Trained model 
+            Trained model
         env : Env
             Gym-like environment
         log_path : str
@@ -147,12 +146,12 @@ class Train_Base():
                         break
             else:
                 log_path += "/test.csv"
-            
+
             with open(log_path, 'w') as f:
                 f.write("reward,ep. length,rew. cumulative avg., ep. len. cumulative avg.\n")
             print("Train statistics are saved to:", log_path)
 
-        if enable_FPS_control: # control simulation speed (using non blocking user input)
+        if enable_FPS_control:  # control simulation speed (using non blocking user input)
             print("\nThe simulation speed can be changed by sending a non-negative integer\n"
                   "(e.g. '50' sets speed to 50%, '0' pauses the simulation, '' sets speed to MAX)\n")
 
@@ -172,8 +171,8 @@ class Train_Base():
             ep_reward += reward
             ep_length += 1
 
-            if enable_FPS_control: # control simulation speed (using non blocking user input)
-                self.control_fps(select.select([sys.stdin], [], [], 0)[0]) 
+            # if enable_FPS_control:  # control simulation speed (using non blocking user input)
+            #     self.control_fps(select.select([sys.stdin], [], [], 0)[0])
 
             if done:
                 obs, _ = env.reset()
@@ -182,25 +181,28 @@ class Train_Base():
                 reward_max = max(ep_reward, reward_max)
                 reward_min = min(ep_reward, reward_min)
                 ep_no += 1
-                avg_ep_lengths = ep_lengths_sum/ep_no
-                avg_rewards = rewards_sum/ep_no
+                avg_ep_lengths = ep_lengths_sum / ep_no
+                avg_rewards = rewards_sum / ep_no
 
                 if verbose > 0:
-                    print(  f"\rEpisode: {ep_no:<3}  Ep.Length: {ep_length:<4.0f}  Reward: {ep_reward:<6.2f}                                                             \n",
-                        end=f"--AVERAGE--   Ep.Length: {avg_ep_lengths:<4.0f}  Reward: {avg_rewards:<6.2f}  (Min: {reward_min:<6.2f}  Max: {reward_max:<6.2f})", flush=True)
-                
+                    print(
+                        f"\rEpisode: {ep_no:<3}  Ep.Length: {ep_length:<4.0f}  Reward: {ep_reward:<6.2f}                                                             \n",
+                        end=f"--AVERAGE--   Ep.Length: {avg_ep_lengths:<4.0f}  Reward: {avg_rewards:<6.2f}  (Min: {reward_min:<6.2f}  Max: {reward_max:<6.2f})",
+                        flush=True)
+
                 if log_path is not None:
                     with open(log_path, 'a') as f:
                         writer = csv.writer(f)
                         writer.writerow([ep_reward, ep_length, avg_rewards, avg_ep_lengths])
-                
+
                 if ep_no == max_episodes:
                     return
 
                 ep_reward = 0
                 ep_length = 0
 
-    def learn_model(self, model:BaseAlgorithm, total_steps:int, path:str, eval_env=None, eval_freq=None, eval_eps=5, save_freq=None, backup_env_file=None, export_name=None):
+    def learn_model(self, model: BaseAlgorithm, total_steps: int, path: str, eval_env=None, eval_freq=None, eval_eps=5,
+                    save_freq=None, backup_env_file=None, export_name=None):
         '''
         Learn Model for a specific number of time steps
 
@@ -251,7 +253,7 @@ class Train_Base():
         # If path already exists, add suffix to avoid overwriting
         if os.path.isdir(path):
             for i in count():
-                p = path.rstrip("/")+f'_{i:03}/'
+                p = path.rstrip("/") + f'_{i:03}/'
                 if not os.path.isdir(p):
                     path = p
                     break
@@ -265,22 +267,28 @@ class Train_Base():
         evaluate = bool(eval_env is not None and eval_freq is not None)
 
         # Create evaluation callback
-        eval_callback = None if not evaluate else EvalCallback(eval_env, n_eval_episodes=eval_eps, eval_freq=eval_freq, log_path=path,
-                                                               best_model_save_path=path, deterministic=True, render=False)
+        eval_callback = None if not evaluate else EvalCallback(eval_env, n_eval_episodes=eval_eps, eval_freq=eval_freq,
+                                                               log_path=path,
+                                                               best_model_save_path=path, deterministic=True,
+                                                               render=False)
 
         # Create custom callback to display evaluations
-        custom_callback = None if not evaluate else Cyclic_Callback(eval_freq, lambda:self.display_evaluations(path,True))
+        custom_callback = None if not evaluate else Cyclic_Callback(eval_freq,
+                                                                    lambda: self.display_evaluations(path, True))
 
         # Create checkpoint callback
-        checkpoint_callback = None if save_freq is None else CheckpointCallback(save_freq=save_freq, save_path=path, name_prefix="model", verbose=1)
+        checkpoint_callback = None if save_freq is None else CheckpointCallback(save_freq=save_freq, save_path=path,
+                                                                                name_prefix="model", verbose=1)
 
         # Create custom callback to export checkpoint models
-        export_callback = None if save_freq is None or export_name is None else Export_Callback(save_freq, path, export_name)
+        export_callback = None if save_freq is None or export_name is None else Export_Callback(save_freq, path,
+                                                                                                export_name)
 
-        callbacks = CallbackList([c for c in [eval_callback, custom_callback, checkpoint_callback, export_callback] if c is not None])
+        callbacks = CallbackList(
+            [c for c in [eval_callback, custom_callback, checkpoint_callback, export_callback] if c is not None])
 
-        model.learn( total_timesteps=total_steps, callback=callbacks )
-        model.save( os.path.join(path, "last_model") )
+        model.learn(total_timesteps=total_steps, callback=callbacks)
+        model.save(os.path.join(path, "last_model"))
 
         # Display evaluations if they exist
         if evaluate:
@@ -288,18 +296,18 @@ class Train_Base():
 
         # Display timestamps + Model path
         end_date = datetime.now().strftime('%d/%m/%Y %H:%M:%S')
-        duration = timedelta(seconds=int(time.time()-start))
+        duration = timedelta(seconds=int(time.time() - start))
         print(f"Train start:     {start_date}")
         print(f"Train end:       {end_date}")
         print(f"Train duration:  {duration}")
         print(f"Model path:      {path}")
-        
+
         # Append timestamps to backup environment file
         if backup_env_file is not None:
             with open(backup_file, 'a') as f:
                 f.write(f"\n# Train start:    {start_date}\n")
-                f.write(  f"# Train end:      {end_date}\n")
-                f.write(  f"# Train duration: {duration}")
+                f.write(f"# Train end:      {end_date}\n")
+                f.write(f"# Train duration: {duration}")
 
         return path
 
@@ -318,50 +326,57 @@ class Train_Base():
 
         with np.load(eval_npz) as data:
             time_steps = data["timesteps"]
-            results_raw = np.mean(data["results"],axis=1)
-            ep_lengths_raw = np.mean(data["ep_lengths"],axis=1)
+            results_raw = np.mean(data["results"], axis=1)
+            ep_lengths_raw = np.mean(data["ep_lengths"], axis=1)
         sample_no = len(results_raw)
 
-        xvals = np.linspace(0, sample_no-1, 80)
-        results    = np.interp(xvals, range(sample_no), results_raw)
+        xvals = np.linspace(0, sample_no - 1, 80)
+        results = np.interp(xvals, range(sample_no), results_raw)
         ep_lengths = np.interp(xvals, range(sample_no), ep_lengths_raw)
 
-        results_limits    = np.min(results),    np.max(results)
+        results_limits = np.min(results), np.max(results)
         ep_lengths_limits = np.min(ep_lengths), np.max(ep_lengths)
 
-        results_discrete    = np.digitize(results,    np.linspace(results_limits[0]-1e-5, results_limits[1]+1e-5,    console_height+1))-1
-        ep_lengths_discrete = np.digitize(ep_lengths, np.linspace(0,                      ep_lengths_limits[1]+1e-5, console_height+1))-1
+        results_discrete = np.digitize(results, np.linspace(results_limits[0] - 1e-5, results_limits[1] + 1e-5,
+                                                            console_height + 1)) - 1
+        ep_lengths_discrete = np.digitize(ep_lengths,
+                                          np.linspace(0, ep_lengths_limits[1] + 1e-5, console_height + 1)) - 1
 
         matrix = np.zeros((console_height, console_width, 2), int)
-        matrix[results_discrete[0]   ][0][0] = 1    # draw 1st column
-        matrix[ep_lengths_discrete[0]][0][1] = 1    # draw 1st column
+        matrix[results_discrete[0]][0][0] = 1  # draw 1st column
+        matrix[ep_lengths_discrete[0]][0][1] = 1  # draw 1st column
         rng = [[results_discrete[0], results_discrete[0]], [ep_lengths_discrete[0], ep_lengths_discrete[0]]]
 
         # Create continuous line for both plots
         for k in range(2):
-            for i in range(1,console_width):
+            for i in range(1, console_width):
                 x = [results_discrete, ep_lengths_discrete][k][i]
                 if x > rng[k][1]:
-                    rng[k] = [rng[k][1]+1, x]
+                    rng[k] = [rng[k][1] + 1, x]
                 elif x < rng[k][0]:
-                    rng[k] = [x, rng[k][0]-1]
+                    rng[k] = [x, rng[k][0] - 1]
                 else:
-                    rng[k] = [x,x]
-                for j in range(rng[k][0],rng[k][1]+1):
+                    rng[k] = [x, x]
+                for j in range(rng[k][0], rng[k][1] + 1):
                     matrix[j][i][k] = 1
 
-        print(f'{"-"*console_width}')
+        print(f'{"-" * console_width}')
         for l in reversed(range(console_height)):
             for c in range(console_width):
-                if   np.all(matrix[l][c] == 0): print(end=" ")
-                elif np.all(matrix[l][c] == 1): print(end=symb_xo)
-                elif matrix[l][c][0] == 1:      print(end=symb_x)
-                else:                           print(end=symb_o)
+                if np.all(matrix[l][c] == 0):
+                    print(end=" ")
+                elif np.all(matrix[l][c] == 1):
+                    print(end=symb_xo)
+                elif matrix[l][c][0] == 1:
+                    print(end=symb_x)
+                else:
+                    print(end=symb_o)
             print()
-        print(f'{"-"*console_width}')
+        print(f'{"-" * console_width}')
         print(f"({symb_x})-reward          min:{results_limits[0]:11.2f}    max:{results_limits[1]:11.2f}")
-        print(f"({symb_o})-ep. length      min:{ep_lengths_limits[0]:11.0f}    max:{ep_lengths_limits[1]:11.0f}    {time_steps[-1]/1000:15.0f}k steps")
-        print(f'{"-"*console_width}')
+        print(
+            f"({symb_o})-ep. length      min:{ep_lengths_limits[0]:11.0f}    max:{ep_lengths_limits[1]:11.0f}    {time_steps[-1] / 1000:15.0f}k steps")
+        print(f'{"-" * console_width}')
 
         # save CSV
         if save_csv:
@@ -370,8 +385,7 @@ class Train_Base():
                 writer = csv.writer(f)
                 if sample_no == 1:
                     writer.writerow(["time_steps", "reward ep.", "length"])
-                writer.writerow([time_steps[-1],results_raw[-1],ep_lengths_raw[-1]])
-
+                writer.writerow([time_steps[-1], results_raw[-1], ep_lengths_raw[-1]])
 
     # def generate_slot_behavior(self, path, slots, auto_head:bool, XML_name):
     #     '''
@@ -394,7 +408,7 @@ class Train_Base():
     #     xml_pretty = minidom.parseString(xml_rough).toprettyxml(indent="  ")
     #     with open(file, "w") as x:
     #         x.write(xml_pretty)
-        
+
     #     print(file, "was created!")
 
     # @staticmethod
@@ -406,7 +420,7 @@ class Train_Base():
     #     ----------
     #     initial_value : float
     #         Initial learning rate
-        
+
     #     Returns
     #     -------
     #     schedule : Callable[[float], float]
@@ -420,7 +434,7 @@ class Train_Base():
     #         ----------
     #         progress_remaining : float
     #             Progress will decrease from 1 (beginning) to 0
-            
+
     #         Returns
     #         -------
     #         learning_rate : float
@@ -452,28 +466,28 @@ class Train_Base():
                 if not os.path.isfile(f):
                     output_file = f
                     break
-        
-        model = PPO.load(input_file)
-        weights = model.policy.state_dict() # dictionary containing network layers
 
-        w = lambda name : weights[name].detach().cpu().numpy() # extract weights from policy
+        model = PPO.load(input_file)
+        weights = model.policy.state_dict()  # dictionary containing network layers
+
+        w = lambda name: weights[name].detach().cpu().numpy()  # extract weights from policy
 
         var_list = []
-        for i in count(0,2): # add hidden layers (step=2 because that's how SB3 works)
+        for i in count(0, 2):  # add hidden layers (step=2 because that's how SB3 works)
             if f"mlp_extractor.policy_net.{i}.bias" not in weights:
                 break
-            var_list.append([w(f"mlp_extractor.policy_net.{i}.bias"), w(f"mlp_extractor.policy_net.{i}.weight"), "tanh"])
+            var_list.append(
+                [w(f"mlp_extractor.policy_net.{i}.bias"), w(f"mlp_extractor.policy_net.{i}.weight"), "tanh"])
 
-        var_list.append( [w("action_net.bias"), w("action_net.weight"), "none"] ) # add final layer
-        
-        with open(output_file,"wb") as f:
-            pickle.dump(var_list, f, protocol=4) # protocol 4 is backward compatible with Python 3.4
+        var_list.append([w("action_net.bias"), w("action_net.weight"), "none"])  # add final layer
+
+        with open(output_file, "wb") as f:
+            pickle.dump(var_list, f, protocol=4)  # protocol 4 is backward compatible with Python 3.4
 
-    
     def print_list(data, numbering=True, prompt=None, divider=" | ", alignment="<", min_per_col=6):
         '''
         Print list - prints list, using as many columns as possible
-        
+
         Parameters
         ----------
         data : `list`
@@ -488,7 +502,7 @@ class Train_Base():
             f-string style alignment ( '<', '>', '^' )
         min_per_col : int
             avoid splitting columns with fewer items
-        
+
         Returns
         -------
         item : `int`, item
@@ -496,65 +510,67 @@ class Train_Base():
             or `None` (if `numbering` or `prompt` are `None`)
 
         '''
-        
+
         WIDTH = shutil.get_terminal_size()[0]
 
-        data_size = len(data)   
+        data_size = len(data)
         items = []
         items_len = []
 
-        #--------------------------------------------- Add numbers, margins and divider
+        # --------------------------------------------- Add numbers, margins and divider
         for i in range(data_size):
             number = f"{i}-" if numbering else ""
-            items.append( f"{divider}{number}{data[i]}" )
-            items_len.append( len(items[-1]) )
+            items.append(f"{divider}{number}{data[i]}")
+            items_len.append(len(items[-1]))
 
-        max_cols = np.clip((WIDTH+len(divider)) // min(items_len),1,math.ceil(data_size/max(min_per_col,1))) # width + len(divider) because it is not needed in last col
+        max_cols = np.clip((WIDTH + len(divider)) // min(items_len), 1, math.ceil(
+            data_size / max(min_per_col, 1)))  # width + len(divider) because it is not needed in last col
 
-        #--------------------------------------------- Check maximum number of columns, considering content width (min:1)
-        for i in range(max_cols,0,-1):
+        # --------------------------------------------- Check maximum number of columns, considering content width (min:1)
+        for i in range(max_cols, 0, -1):
             cols_width = []
             cols_items = []
             table_width = 0
-            a,b = divmod(data_size,i)
+            a, b = divmod(data_size, i)
             for col in range(i):
-                start = a*col + min(b,col)
-                end = start+a+(1 if col<b else 0)
-                cols_items.append( items[start:end] )
+                start = a * col + min(b, col)
+                end = start + a + (1 if col < b else 0)
+                cols_items.append(items[start:end])
                 col_width = max(items_len[start:end])
-                cols_width.append( col_width )
+                cols_width.append(col_width)
                 table_width += col_width
-            if table_width <= WIDTH+len(divider):
+            if table_width <= WIDTH + len(divider):
                 break
         table_width -= len(divider)
-        
-        #--------------------------------------------- Print columns
-        print("="*table_width)
+
+        # --------------------------------------------- Print columns
+        print("=" * table_width)
         for row in range(math.ceil(data_size / i)):
             for col in range(i):
-                content = cols_items[col][row] if len(cols_items[col]) > row else divider # print divider when there are no items
+                content = cols_items[col][row] if len(
+                    cols_items[col]) > row else divider  # print divider when there are no items
                 if col == 0:
                     l = len(divider)
-                    print(end=f"{content[l:]:{alignment}{cols_width[col]-l}}")  # remove divider from 1st col
+                    print(end=f"{content[l:]:{alignment}{cols_width[col] - l}}")  # remove divider from 1st col
                 else:
-                    print(end=f"{content    :{alignment}{cols_width[col]  }}") 
-            print()  
-        print("="*table_width)
+                    print(end=f"{content    :{alignment}{cols_width[col]}}")
+            print()
+        print("=" * table_width)
 
-        #--------------------------------------------- Prompt
+        # --------------------------------------------- Prompt
         if prompt is None:
             return None
 
         if numbering is None:
             return None
         else:
-            idx = UI.read_int( prompt, 0, data_size )
+            idx = UI.read_int(prompt, 0, data_size)
             return idx, data[idx]
 
 
-
 class Cyclic_Callback(BaseCallback):
     ''' Stable baselines custom callback '''
+
     def __init__(self, freq, function):
         super(Cyclic_Callback, self).__init__(1)
         self.freq = freq
@@ -563,10 +579,12 @@ class Cyclic_Callback(BaseCallback):
     def _on_step(self) -> bool:
         if self.n_calls % self.freq == 0:
             self.function()
-        return True # If the callback returns False, training is aborted early
+        return True  # If the callback returns False, training is aborted early
+
 
 class Export_Callback(BaseCallback):
     ''' Stable baselines custom callback '''
+
     def __init__(self, freq, load_path, export_name):
         super(Export_Callback, self).__init__(1)
         self.freq = freq
@@ -577,8 +595,7 @@ class Export_Callback(BaseCallback):
         if self.n_calls % self.freq == 0:
             path = os.path.join(self.load_path, f"model_{self.num_timesteps}_steps.zip")
             Train_Base.export_model(path, f"./scripts/gyms/export/{self.export_name}")
-        return True # If the callback returns False, training is aborted early
-    
+        return True  # If the callback returns False, training is aborted early
+
 
 
-    

scripts/gyms/Walk.py

@@ -1,13 +1,15 @@
 import os
 import numpy as np
 import math
-from time import sleep 
+import time
+from time import sleep
 from random import random
 from random import uniform
-
+from itertools import count
 
 from stable_baselines3 import PPO
-from stable_baselines3.common.vec_env import SubprocVecEnv
+from stable_baselines3.common.monitor import Monitor
+from stable_baselines3.common.vec_env import SubprocVecEnv, DummyVecEnv
 
 import gymnasium as gym
 from gymnasium import spaces
@@ -28,56 +30,65 @@ Learn how to run forward using step primitive
 - class Train:  implements algorithms to train a new model or test an existing model
 '''
 
+
 class WalkEnv(gym.Env):
     def __init__(self, ip, server_p) -> None:
 
-        
-
         # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
         self.Player = player = Base_Agent(
-                team_name="Gym",
-                number=1,
-                host=ip,
-                port=server_p
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
         )
-        self.robot_type = self.Player.robot       
-        self.step_counter = 0 # to limit episode size
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
         self.force_play_on = True
 
-        self.target_position = np.array([0.0, 0.0]) # target position in the x-y plane 
-        self.initial_position = np.array([0.0, 0.0]) # initial position in the x-y plane
-        self.target_direction = 0.0 # target direction in the x-y plane (relative to the robot's orientation)
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
         self.isfallen = False
         self.waypoint_index = 0
         self.route_completed = False
         self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
         self.calibrate_nominal_from_neutral = True
         self.auto_calibrate_train_sim_flip = True
         self.nominal_calibrated_once = False
         self.flip_calibrated_once = False
-
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
 
         # State space
         # 原始观测大小: 78
         obs_size = 78
         self.obs = np.zeros(obs_size, np.float32)
         self.observation_space = spaces.Box(
-            low=-10.0, 
-            high=10.0, 
-            shape=(obs_size,), 
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
             dtype=np.float32
         )
 
         action_dim = len(self.Player.robot.ROBOT_MOTORS)
         self.no_of_actions = action_dim
         self.action_space = spaces.Box(
-            low=-1.0,
-            high=1.0,
-            shape=(action_dim,), 
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
             dtype=np.float32
         )
 
-        
         # 中立姿态
         self.joint_nominal_position = np.array(
             [
@@ -106,26 +117,25 @@ class WalkEnv(gym.Env):
                 0.0,
             ]
         )
-        self.reference_joint_nominal_position = self.joint_nominal_position.copy()
-
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
         self.train_sim_flip = np.array(
             [
-                1.0,   # 0: Head_yaw (he1)
+                1.0,  # 0: Head_yaw (he1)
                 -1.0,  # 1: Head_pitch (he2)
-                1.0,   # 2: Left_Shoulder_Pitch (lae1)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
                 -1.0,  # 3: Left_Shoulder_Roll (lae2)
-                1.0,   # 4: Left_Elbow_Pitch (lae3)
-                1.0,   # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
                 -1.0,  # 6: Right_Shoulder_Pitch (rae1)
-                1.0,   # 7: Right_Shoulder_Roll (rae2)
-                1.0,   # 8: Right_Elbow_Pitch (rae3)
-                1.0,   # 9: Right_Elbow_Yaw (rae4)
-                1.0,   # 10: Waist (te1)
-                1.0,   # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
                 -1.0,  # 12: Left_Hip_Roll (lle2)
                 -1.0,  # 13: Left_Hip_Yaw (lle3)
-                1.0,   # 14: Left_Knee_Pitch (lle4)
-                1.0,   # 15: Left_Ankle_Pitch (lle5)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
                 -1.0,  # 16: Left_Ankle_Roll (lle6)
                 -1.0,  # 17: Right_Hip_Pitch (rle1)
                 -1.0,  # 18: Right_Hip_Roll (rle2)
@@ -136,15 +146,54 @@ class WalkEnv(gym.Env):
             ]
         )
 
-        self.scaling_factor = 0.5
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Encourage a minimum lateral stance so the policy avoids feet overlap.
+        self.min_stance_rad = 0.10
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = True
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.025
+        self.reset_perturb_steps = 4
+        self.reset_recover_steps = 8
 
         self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
         self.previous_pos = np.array([0.0, 0.0])  # Track previous position
         self.Player.server.connect()
-        sleep(2.0)  # Longer wait for connection to establish completely
+        # sleep(2.0)  # Longer wait for connection to establish completely
         self.Player.server.send_immediate(
             f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
         )
+        self.start_time = time.time()
+
+    def _reconnect_server(self):
+        try:
+            self.Player.server.shutdown()
+        except Exception:
+            pass
+
+        self.Player.server.connect()
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+
+    def _safe_receive_world_update(self, retries=1):
+        last_exc = None
+        for attempt in range(retries + 1):
+            try:
+                self.Player.server.receive()
+                self.Player.world.update()
+                return
+            except (ConnectionResetError, OSError) as exc:
+                last_exc = exc
+                if attempt >= retries:
+                    raise
+                self._reconnect_server()
+        if last_exc is not None:
+            raise last_exc
 
     def debug_log(self, message):
         print(message)
@@ -155,70 +204,30 @@ class WalkEnv(gym.Env):
         except OSError:
             pass
 
-    def calibrate_train_sim_flip_from_neutral(self, neutral_joint_positions):
-        updated_flip = self.train_sim_flip.copy()
-        changed = []
-
-        for idx, (reference_value, observed_value) in enumerate(
-            zip(self.reference_joint_nominal_position, neutral_joint_positions)
-        ):
-            if idx >= 10:
-                continue
-            if abs(reference_value) < 0.15 or abs(observed_value) < 0.15:
-                continue
-
-            inferred_flip = 1.0 if np.sign(reference_value) == np.sign(observed_value) else -1.0
-            if updated_flip[idx] != inferred_flip:
-                changed.append((idx, updated_flip[idx], inferred_flip))
-                updated_flip[idx] = inferred_flip
-
-        self.train_sim_flip = updated_flip
-
-        if changed:
-            self.debug_log(
-                "[FlipDebug] "
-                f"changes={[(idx, old, new) for idx, old, new in changed]}"
-            )
-
-    def is_reliable_neutral_pose(self, neutral_joint_positions):
-        leg_positions = neutral_joint_positions[11:]
-        leg_norm = float(np.linalg.norm(leg_positions))
-        leg_max = float(np.max(np.abs(leg_positions)))
-        height = float(self.Player.world.global_position[2])
-
-        reliable = (
-            leg_norm > 0.8
-            and leg_max > 0.35
-            and 0.12 < height < 0.8
-        )
-
-        return reliable, leg_norm, leg_max, height
-
-
     def observe(self, init=False):
 
         """获取当前观测值"""
         robot = self.Player.robot
         world = self.Player.world
-        
+
         # Safety check: ensure data is available
-        
+
         # 计算目标速度
         raw_target = self.target_position - world.global_position[:2]
         velocity = MathOps.rotate_2d_vec(
-            raw_target, 
-            -robot.global_orientation_euler[2], 
+            raw_target,
+            -robot.global_orientation_euler[2],
             is_rad=False
         )
-        
+
         # 计算相对方向
         rel_orientation = MathOps.vector_angle(velocity) * 0.3
         rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
-        
+
         velocity = np.concatenate([velocity, np.array([rel_orientation])])
         velocity[0] = np.clip(velocity[0], -0.5, 0.5)
         velocity[1] = np.clip(velocity[1], -0.25, 0.25)
-        
+
         # 关节状态
         radian_joint_positions = np.deg2rad(
             [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
@@ -226,21 +235,20 @@ class WalkEnv(gym.Env):
         radian_joint_speeds = np.deg2rad(
             [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
         )
-        
+
         qpos_qvel_previous_action = np.concatenate([
             (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
             radian_joint_speeds / 110.0 * self.train_sim_flip,
             self.previous_action / 10.0,
         ])
-        
+
         # 角速度
         ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
-        
+
         # 投影的重力方向
         orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
         projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
-        
-        
+
         # 组合观测
         observation = np.concatenate([
             qpos_qvel_previous_action,
@@ -249,17 +257,25 @@ class WalkEnv(gym.Env):
             projected_gravity,
         ])
 
-        
-        
         observation = np.clip(observation, -10.0, 10.0)
         return observation.astype(np.float32)
 
     def sync(self):
         ''' Run a single simulation step '''
-        self.Player.server.receive()
-        self.Player.world.update()
+        self._safe_receive_world_update(retries=1)
         self.Player.robot.commit_motor_targets_pd()
         self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
 
     def debug_joint_status(self):
         robot = self.Player.robot
@@ -284,6 +300,7 @@ class WalkEnv(gym.Env):
             f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
             f"fallen={self.Player.world.global_position[2] < 0.3}"
         )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
 
     def reset(self, seed=None, options=None):
         '''
@@ -295,83 +312,82 @@ class WalkEnv(gym.Env):
         if seed is not None:
             np.random.seed(seed)
 
-        length1 = np.random.uniform(10, 20) # randomize target distance
-        length2 = np.random.uniform(10, 20) # randomize target distance
-        length3 = np.random.uniform(10, 20) # randomize target distance
-        angle2 = np.random.uniform(-30, 30) # randomize initial orientation
-        angle3 = np.random.uniform(-30, 30) # randomize target direction
-
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
 
         self.step_counter = 0
         self.waypoint_index = 0
         self.route_completed = False
         self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
         self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
         self.walk_cycle_step = 0
 
         # 随机 beam 目标位置和朝向，增加训练多样性
         beam_x = (random() - 0.5) * 10
         beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
 
         for _ in range(5):
-            self.Player.server.receive()
-            self.Player.world.update()
+            self._safe_receive_world_update(retries=2)
             self.Player.robot.commit_motor_targets_pd()
-            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=0)
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
             self.Player.server.send()
 
         # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
         finished_count = 0
-        for _ in range(20):
+        for _ in range(50):
             finished = self.Player.skills_manager.execute("Neutral")
             self.sync()
             if finished:
                 finished_count += 1
-                if finished_count >= 2:  # 假设需要连续2次完成才算成功
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
                     break
 
-        # neutral_joint_positions = np.deg2rad(
-        #     [self.Player.robot.motor_positions[motor] for motor in self.Player.robot.ROBOT_MOTORS]
-        # )
-        # reliable_neutral, neutral_leg_norm, neutral_leg_max, neutral_height = self.is_reliable_neutral_pose(neutral_joint_positions)
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
 
-        # if self.auto_calibrate_train_sim_flip and reliable_neutral and not self.flip_calibrated_once:
-        #     self.calibrate_train_sim_flip_from_neutral(neutral_joint_positions)
-        #     self.flip_calibrated_once = True
-        # if self.calibrate_nominal_from_neutral and reliable_neutral and not self.nominal_calibrated_once:
-        #     self.joint_nominal_position = neutral_joint_positions * self.train_sim_flip
-        #     self.nominal_calibrated_once = True
-        # self.debug_log(
-        #     "[ResetDebug] "
-        #     f"neutral_pos={np.round(self.Player.world.global_position, 3).tolist()} "
-        #     f"shoulders={np.round(neutral_joint_positions[2:10], 3).tolist()} "
-        #     f"legs={np.round(neutral_joint_positions[11:], 3).tolist()} "
-        #     f"flip={self.train_sim_flip.tolist()} "
-        #     f"nominal_legs={np.round(self.joint_nominal_position[11:], 3).tolist()} "
-        #     f"calibrated_once={(self.flip_calibrated_once, self.nominal_calibrated_once)} "
-        #     f"reliable_neutral={reliable_neutral} "
-        #     f"leg_norm={neutral_leg_norm:.3f} leg_max={neutral_leg_max:.3f} height={neutral_height:.3f}"
-        # )
-
-        # reset_action_noise = np.random.uniform(-0.015, 0.015, size=(len(self.Player.robot.ROBOT_MOTORS),))
-        # self.target_joint_positions = (self.joint_nominal_position + reset_action_noise) * self.train_sim_flip
-
-
-        # for idx, target in enumerate(self.target_joint_positions):
-        #     r.set_motor_target_position(
-        #         r.ROBOT_MOTORS[idx], target*180/math.pi, kp=25, kd=0.6
-        #     )
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
 
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
 
         # memory variables
+        self.sync()  
         self.initial_position = np.array(self.Player.world.global_position[:2])
         self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
-        self.act = np.zeros(self.no_of_actions ,np.float32)
-        point1 = self.initial_position + np.array([length1, 0])
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
         point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
         point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
-        self.point_list = [point1, point2, point3]
+        self.point_list = [point1]
         self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
 
         return self.observe(True), {}
 
@@ -379,101 +395,158 @@ class WalkEnv(gym.Env):
         return
 
     def compute_reward(self, previous_pos, current_pos, action):
-        velocity = current_pos - previous_pos
-        velocity_magnitude = np.linalg.norm(velocity)
-        direction_to_target = self.target_position - current_pos
-        prev_direction_to_target = self.target_position - previous_pos
-        distance_to_target = np.linalg.norm(direction_to_target)
-        prev_distance_to_target = np.linalg.norm(prev_direction_to_target)
+        height = float(self.Player.world.global_position[2])
+        robot = self.Player.robot
 
-        progress_reward = np.clip((prev_distance_to_target - distance_to_target) * 30.0, -2.0, 4.0)
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
 
-        velocity_in_m_per_sec = velocity_magnitude / 0.05
-        speed_reward = np.clip(velocity_in_m_per_sec * 1.5, 0.0, 1.5)
+        is_fallen = height < 0.55
+        if is_fallen:
+            # remain = max(0, 800 - self.step_counter)
+            # return -8.0 - 0.01 * remain
+            return -1.0
+        
 
-        if velocity_magnitude > 1e-4 and distance_to_target > 1e-4:
-            directional_alignment = np.dot(velocity, direction_to_target) / (velocity_magnitude * distance_to_target)
-            directional_alignment = np.clip(directional_alignment, -1.0, 1.0)
-            direction_reward = max(0.0, directional_alignment)
-        else:
-            direction_reward = 0.0
 
-        alive_bonus = 0.05
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
 
-        height = self.Player.world.global_position[2]
-        if 0.45 <= height <= 1.2:
-            height_reward = 1.5
-        else:
-            height_reward = -6.0
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
 
-        motionless_penalty = -1.5 if velocity_magnitude < 0.003 else 0.0
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 2.0
 
-        waypoint_bonus = 0.0
-        if distance_to_target < 0.5:
-            waypoint_bonus = 25.0
-            if self.waypoint_index < len(self.point_list) - 1:
-                self.waypoint_index += 1
-                self.target_position = self.point_list[self.waypoint_index]
-            else:
-                waypoint_bonus = 100.0
-                self.route_completed = True
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.01 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -0.3 * (tilt_mag)
+        ang_vel_penalty = -0.02 * ang_vel_mag
+
+        # Use simulator joint readings in training frame to shape lateral stance.
+        joint_pos = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        ) * self.train_sim_flip
+        left_hip_roll = float(joint_pos[12])
+        right_hip_roll = float(joint_pos[18])
+        left_ankle_roll = float(joint_pos[16])
+        right_ankle_roll = float(joint_pos[22])
+
+        hip_spread = left_hip_roll - right_hip_roll
+        ankle_spread = left_ankle_roll - right_ankle_roll
+        stance_metric = 0.6 * abs(hip_spread) + 0.4 * abs(ankle_spread)
+
+        # Penalize narrow stance (feet too close) and scissoring (cross-leg pattern).
+        stance_collapse_penalty = -4.0 * max(0.0, self.min_stance_rad - stance_metric)
+        cross_leg_penalty = -1.2 * max(0.0, -(hip_spread * ankle_spread))
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -0.5 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                 + stance_collapse_penalty
+                 + cross_leg_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 600:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                  f"stance_collapse_penalty:{stance_collapse_penalty:.4f}",
+                  f"cross_leg_penalty:{cross_leg_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
 
-        action_magnitude = np.linalg.norm(action[11:])
-        action_penalty = -0.08 * action_magnitude
-        tilt_penalty = -0.2 * np.linalg.norm(self.Player.robot.gyroscope[:2]) / 100.0
 
-        return (
-            progress_reward
-            + speed_reward
-            + direction_reward
-            + alive_bonus
-            + height_reward
-            + motionless_penalty
-            + waypoint_bonus
-            + action_penalty
-            + tilt_penalty
-        )
 
     def step(self, action):
-        
+
         r = self.Player.robot
-
-
         self.previous_action = action
 
         self.target_joint_positions = (
-            self.joint_nominal_position
-            + self.scaling_factor * action
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
         )
         self.target_joint_positions *= self.train_sim_flip
 
-
         for idx, target in enumerate(self.target_joint_positions):
             r.set_motor_target_position(
-                r.ROBOT_MOTORS[idx], target*180/math.pi, kp=25, kd=0.6
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=40, kd=1.0
             )
-        
-        
 
-        
-        self.sync() # run simulation step
+        self.previous_action = action
+
+        self.sync()  # run simulation step
         self.step_counter += 1
 
-        # if self.step_counter % self.debug_every_n_steps == 0:
-        #     self.debug_joint_status()
-         
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
         current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
-        
+
         # Compute reward based on movement from previous step
         reward = self.compute_reward(self.previous_pos, current_pos, action)
-        
-        
+
         # Update previous position
         self.previous_pos = current_pos.copy()
-        
+        self.last_action_for_reward = action.copy()
+
         # Fall detection and penalty
-        is_fallen = self.Player.world.global_position[2] < 0.3
-        
+        is_fallen = self.Player.world.global_position[2] < 0.55
+
         # terminal state: the robot is falling or timeout
         terminated = is_fallen or self.step_counter > 800 or self.route_completed
         truncated = False
@@ -481,94 +554,119 @@ class WalkEnv(gym.Env):
         return self.observe(), reward, terminated, truncated, {}
 
 
-
-
-
 class Train(Train_Base):
     def __init__(self, script) -> None:
         super().__init__(script)
 
-
     def train(self, args):
 
-        #--------------------------------------- Learning parameters
-        n_envs = 8  # Reduced from 8 to decrease CPU/network pressure during init
-        n_steps_per_env = 512  # RolloutBuffer is of size (n_steps_per_env * n_envs)
-        minibatch_size = 64    # should be a factor of (n_steps_per_env * n_envs)
+        # --------------------------------------- Learning parameters
+        n_envs = int(os.environ.get("GYM_CPU_N_ENVS", "20"))
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        server_warmup_sec = float(os.environ.get("GYM_CPU_SERVER_WARMUP_SEC", "3.0"))
+        n_steps_per_env = int(os.environ.get("GYM_CPU_TRAIN_STEPS_PER_ENV", "256"))  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = int(os.environ.get("GYM_CPU_TRAIN_BATCH_SIZE", "512"))  # should be a factor of (n_steps_per_env * n_envs)
         total_steps = 30000000
-        learning_rate = 3e-4
+        learning_rate = float(os.environ.get("GYM_CPU_TRAIN_LR", "3e-4"))
         folder_name = f'Walk_R{self.robot_type}'
-        model_path = f'./scripts/gyms/logs/{folder_name}/' 
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
 
         print(f"Model path: {model_path}")
         print(f"Using {n_envs} parallel environments")
 
-        #--------------------------------------- Run algorithm
-        def init_env(i_env):
+        # --------------------------------------- Run algorithm
+        def init_env(i_env, monitor=False):
             def thunk():
-                return WalkEnv( self.ip , self.server_p + i_env)
+                env = WalkEnv(self.ip, self.server_p + i_env)
+                if monitor:
+                    env = Monitor(env)
+                return env
+
             return thunk
 
-        servers = Train_Server( self.server_p, self.monitor_p_1000, n_envs+1 ) #include 1 extra server for testing
-        
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1, no_render=True, no_realtime=True)  # include 1 extra server for testing
+
         # Wait for servers to start
-        print(f"Starting {n_envs+1} rcssservermj servers...")
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        if server_warmup_sec > 0:
+            print(f"Waiting {server_warmup_sec:.1f}s for server warmup...")
+            sleep(server_warmup_sec)
         print("Servers started, creating environments...")
 
-        env = SubprocVecEnv( [init_env(i) for i in range(n_envs)] )
-        eval_env = SubprocVecEnv( [init_env(n_envs)] )
-        
+        env = SubprocVecEnv([init_env(i, monitor=True) for i in range(n_envs)])
+        # Use single-process eval env to avoid extra subprocess fragility during callback evaluation.
+        eval_env = DummyVecEnv([init_env(n_envs, monitor=True)])
 
         try:
             # Custom policy network architecture
             policy_kwargs = dict(
                 net_arch=dict(
-                    pi=[256, 256, 128],  # Policy network: 3 layers
-                    vf=[256, 256, 128]   # Value network: 3 layers  
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
                 ),
-                activation_fn=__import__('torch.nn', fromlist=['ReLU']).ReLU,
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
             )
-            
-            if "model_file" in args: # retrain
-                model = PPO.load( args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env, batch_size=minibatch_size, learning_rate=learning_rate )
-            else: # train new model
-                model = PPO( 
-                    "MlpPolicy", 
-                    env=env, 
-                    verbose=1, 
-                    n_steps=n_steps_per_env, 
-                    batch_size=minibatch_size, 
-                    learning_rate=learning_rate, 
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
                     device="cpu",
                     policy_kwargs=policy_kwargs,
-                    ent_coef=0.01,  # Entropy coefficient for exploration
-                    clip_range=0.2,  # PPO clipping parameter
+                    ent_coef=float(os.environ.get("GYM_CPU_TRAIN_ENT_COEF", "0.05")),  # Entropy coefficient for exploration
+                    clip_range=float(os.environ.get("GYM_CPU_TRAIN_CLIP_RANGE", "0.2")),  # PPO clipping parameter
                     gae_lambda=0.95,  # GAE lambda
-                    gamma=0.99  # Discount factor
+                    gamma=float(os.environ.get("GYM_CPU_TRAIN_GAMMA", "0.95")), # Discount factor
+                    # target_kl=0.03,
+                    n_epochs=int(os.environ.get("GYM_CPU_TRAIN_EPOCHS", "5")),
+                    tensorboard_log=f"./scripts/gyms/logs/{folder_name}/tensorboard/"
                 )
 
-            model_path = self.learn_model( model, total_steps, model_path, eval_env=eval_env, eval_freq=n_steps_per_env*20, save_freq=n_steps_per_env*20, backup_env_file=__file__ )
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 20, save_freq=n_steps_per_env * 20, eval_eps=100,
+                                          backup_env_file=__file__)
         except KeyboardInterrupt:
-            sleep(1) # wait for child processes
+            sleep(1)  # wait for child processes
             print("\nctrl+c pressed, aborting...\n")
             servers.kill()
             return
-    
+
         env.close()
         eval_env.close()
         servers.kill()
-        
 
     def test(self, args):
 
         # Uses different server and monitor ports
-        server = Train_Server( self.server_p-1, self.monitor_p, 1 )
-        env = WalkEnv( self.ip, self.server_p-1 )
-        model = PPO.load( args["model_file"], env=env )
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        test_no_render = os.environ.get("GYM_CPU_TEST_NO_RENDER", "0") == "1"
+        test_no_realtime = os.environ.get("GYM_CPU_TEST_NO_REALTIME", "0") == "1"
+
+        server = Train_Server(
+            self.server_p - 1,
+            self.monitor_p,
+            1,
+            no_render=test_no_render,
+            no_realtime=test_no_realtime,
+        )
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
 
         try:
-            self.export_model( args["model_file"], args["model_file"]+".pkl", False )  # Export to pkl to create custom behavior
-            self.test_model( model, env, log_path=args["folder_dir"], model_path=args["folder_dir"] )
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
         except KeyboardInterrupt:
             print()
 
@@ -578,18 +676,30 @@ class Train(Train_Base):
 
 if __name__ == "__main__":
     from types import SimpleNamespace
-    
+
     # 创建默认参数
     script_args = SimpleNamespace(
         args=SimpleNamespace(
             i='127.0.0.1',  # Server IP
-            p=3100,         # Server port
-            m=3200,         # Monitor port
-            r=0,            # Robot type
-            t='Gym',        # Team name
-            u=1             # Uniform number
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
         )
     )
-    
+
     trainer = Train(script_args)
-    trainer.train({})
+
+    run_mode = os.environ.get("GYM_CPU_MODE", "train").strip().lower()
+
+    if run_mode == "test":
+        test_model_file = os.environ.get("GYM_CPU_TEST_MODEL", "scripts/gyms/logs/Walk_R0_004/best_model.zip")
+        test_folder = os.environ.get("GYM_CPU_TEST_FOLDER", "scripts/gyms/logs/Walk_R0_004/")
+        trainer.test({"model_file": test_model_file, "folder_dir": test_folder})
+    else:
+        retrain_model = os.environ.get("GYM_CPU_TRAIN_MODEL", "").strip()
+        if retrain_model:
+            trainer.train({"model_file": retrain_model})
+        else:
+            trainer.train({})

scripts/gyms/logs/Walk_R0_000/Walk.py

@@ -0,0 +1,624 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.vec_env import SubprocVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-1.0,
+            high=1.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.5
+        # self.scaling_factor = 1
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = False
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.035
+        self.reset_perturb_steps = 5
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self.Player.server.receive()
+        self.Player.world.update()
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self.Player.server.receive()
+            self.Player.world.update()
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+
+        orientation_quat_inv = R.from_quat(self.Player.robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(self.Player.robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        # 摔倒检测（重要！）
+        if height < 0.3:
+            if time.time() - self.start_time > 1200:
+                self.start_time = time.time()
+                print("fall_penalty: -20")
+            return -20.0
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 0.1
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.05 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -1.0 * (tilt_mag)
+        # ang_vel_penalty = -0.05 * ang_vel_mag
+
+        target_height = self.initial_height
+        height_error =  height - target_height if abs(height - target_height) > 0.05 else 0.0
+        height_penalty = -2.0 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                #  + ang_vel_penalty
+                 + height_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 1200:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                self.joint_nominal_position
+                + self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        # self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = 20  # Reduced from 8 to decrease CPU/network pressure during init
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        n_steps_per_env = 1024  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = 128  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = 3e-4
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env):
+            def thunk():
+                return WalkEnv(self.ip, self.server_p + i_env)
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i) for i in range(n_envs)])
+        eval_env = SubprocVecEnv([init_env(n_envs)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=0.005,  # Entropy coefficient for exploration
+                    # clip_range=0.13,  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=0.99 , # Discount factor
+                    # target_kl=0.03,
+                    # n_epochs=5
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 10, save_freq=n_steps_per_env * 10,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        server = Train_Server(self.server_p - 1, self.monitor_p, 1)
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+    trainer.train({})
+    # trainer.test({"model_file": "scripts/gyms/logs/Walk_R0_012/best_model.zip",
+                #   "folder_dir": "scripts/gyms/logs/Walk_R0_012/",})

scripts/gyms/logs/Walk_R0_001/Walk.py

@@ -0,0 +1,625 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.vec_env import SubprocVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.1
+        # self.scaling_factor = 1
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = False
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.035
+        self.reset_perturb_steps = 5
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self.Player.server.receive()
+        self.Player.world.update()
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self.Player.server.receive()
+            self.Player.world.update()
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+
+        orientation_quat_inv = R.from_quat(self.Player.robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(self.Player.robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.3
+        if is_fallen:
+            remain = max(0, 800 - self.step_counter)
+            return -8.0 - 0.01 * remain
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 0.3
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.03 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -1.0 * (tilt_mag)
+        # ang_vel_penalty = -0.05 * ang_vel_mag
+
+        target_height = self.initial_height
+        height_error =  height - target_height if abs(height - target_height) > 0.05 else 0.0
+        height_penalty = -2.0 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                #  + ang_vel_penalty
+                 + height_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 1200:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = 20  # Reduced from 8 to decrease CPU/network pressure during init
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        n_steps_per_env = 1024  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = 128  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = 3e-4
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env):
+            def thunk():
+                return WalkEnv(self.ip, self.server_p + i_env)
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i) for i in range(n_envs)])
+        eval_env = SubprocVecEnv([init_env(n_envs)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=0.005,  # Entropy coefficient for exploration
+                    # clip_range=0.13,  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=0.99 , # Discount factor
+                    # target_kl=0.03,
+                    # n_epochs=5
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 10, save_freq=n_steps_per_env * 10,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        server = Train_Server(self.server_p - 1, self.monitor_p, 1)
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+    trainer.train({})
+    # trainer.test({"model_file": "scripts/gyms/logs/Walk_R0_000/best_model.zip",
+                #   "folder_dir": "scripts/gyms/logs/Walk_R0_000/",})

scripts/gyms/logs/Walk_R0_002/Walk.py

@@ -0,0 +1,625 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.vec_env import SubprocVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.1
+        # self.scaling_factor = 1
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = False
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.035
+        self.reset_perturb_steps = 5
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self.Player.server.receive()
+        self.Player.world.update()
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self.Player.server.receive()
+            self.Player.world.update()
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+
+        orientation_quat_inv = R.from_quat(self.Player.robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(self.Player.robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.3
+        if is_fallen:
+            remain = max(0, 800 - self.step_counter)
+            return -8.0 - 0.01 * remain
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 0.3
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.03 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -1.0 * (tilt_mag)
+        # ang_vel_penalty = -0.05 * ang_vel_mag
+
+        target_height = self.initial_height
+        height_error =  height - target_height if abs(height - target_height) > 0.05 else 0.0
+        height_penalty = -2.0 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                #  + ang_vel_penalty
+                 + height_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 1200:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = 20  # Reduced from 8 to decrease CPU/network pressure during init
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        n_steps_per_env = 1024  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = 128  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = 3e-4
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env):
+            def thunk():
+                return WalkEnv(self.ip, self.server_p + i_env)
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i) for i in range(n_envs)])
+        eval_env = SubprocVecEnv([init_env(n_envs)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=0.005,  # Entropy coefficient for exploration
+                    # clip_range=0.13,  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=0.99 , # Discount factor
+                    # target_kl=0.03,
+                    # n_epochs=5
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 10, save_freq=n_steps_per_env * 10,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        server = Train_Server(self.server_p - 1, self.monitor_p, 1)
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+    trainer.train({})
+    # trainer.test({"model_file": "scripts/gyms/logs/Walk_R0_000/best_model.zip",
+                #   "folder_dir": "scripts/gyms/logs/Walk_R0_000/",})

scripts/gyms/logs/Walk_R0_003/Walk.py

@@ -0,0 +1,625 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.vec_env import SubprocVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = True
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.035
+        self.reset_perturb_steps = 5
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self.Player.server.receive()
+        self.Player.world.update()
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self.Player.server.receive()
+            self.Player.world.update()
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+
+        orientation_quat_inv = R.from_quat(self.Player.robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(self.Player.robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.3
+        if is_fallen:
+            remain = max(0, 800 - self.step_counter)
+            return -8.0 - 0.01 * remain
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 0.3
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.03 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -1.0 * (tilt_mag)
+        ang_vel_penalty = -0.05 * ang_vel_mag
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -2.0 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 1200:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = 20  # Reduced from 8 to decrease CPU/network pressure during init
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        n_steps_per_env = 1024  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = 128  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = 1e-4
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env):
+            def thunk():
+                return WalkEnv(self.ip, self.server_p + i_env)
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i) for i in range(n_envs)])
+        eval_env = SubprocVecEnv([init_env(n_envs)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=0.001,  # Entropy coefficient for exploration
+                    # clip_range=0.13,  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=0.99 , # Discount factor
+                    target_kl=0.03,
+                    # n_epochs=5
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 10, save_freq=n_steps_per_env * 10,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        server = Train_Server(self.server_p - 1, self.monitor_p, 1)
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+    trainer.train({})
+    # trainer.test({"model_file": "scripts/gyms/logs/Walk_R0_000/best_model.zip",
+                #   "folder_dir": "scripts/gyms/logs/Walk_R0_000/",})

scripts/gyms/logs/Walk_R0_004/Walk.py

@@ -0,0 +1,626 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.vec_env import SubprocVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = True
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.015
+        self.reset_perturb_steps = 3
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self.Player.server.receive()
+        self.Player.world.update()
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self.Player.server.receive()
+            self.Player.world.update()
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+
+        orientation_quat_inv = R.from_quat(self.Player.robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(self.Player.robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.3
+        if is_fallen:
+            # remain = max(0, 800 - self.step_counter)
+            # return -8.0 - 0.01 * remain
+            return -1.0
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 2.0
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.01 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -0.3 * (tilt_mag)
+        ang_vel_penalty = -0.02 * ang_vel_mag
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -0.5 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 1200:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=40, kd=1.0
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = 20  # Reduced from 8 to decrease CPU/network pressure during init
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        n_steps_per_env = 256  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = 512  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = 3e-4
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env):
+            def thunk():
+                return WalkEnv(self.ip, self.server_p + i_env)
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i) for i in range(n_envs)])
+        eval_env = SubprocVecEnv([init_env(n_envs)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=0.05,  # Entropy coefficient for exploration
+                    # clip_range=0.13,  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=0.95 , # Discount factor
+                    target_kl=0.03,
+                    n_epochs=5
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 10, save_freq=n_steps_per_env * 10,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        server = Train_Server(self.server_p - 1, self.monitor_p, 1)
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+    trainer.train({})
+    # trainer.test({"model_file": "scripts/gyms/logs/Walk_R0_000/best_model.zip",
+                #   "folder_dir": "scripts/gyms/logs/Walk_R0_000/",})

scripts/gyms/logs/Walk_R0_005/Walk.py

@@ -0,0 +1,660 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.monitor import Monitor
+from stable_baselines3.common.vec_env import SubprocVecEnv, DummyVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = True
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.015
+        self.reset_perturb_steps = 3
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def _reconnect_server(self):
+        try:
+            self.Player.server.shutdown()
+        except Exception:
+            pass
+
+        self.Player.server.connect()
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+
+    def _safe_receive_world_update(self, retries=1):
+        last_exc = None
+        for attempt in range(retries + 1):
+            try:
+                self.Player.server.receive()
+                self.Player.world.update()
+                return
+            except (ConnectionResetError, OSError) as exc:
+                last_exc = exc
+                if attempt >= retries:
+                    raise
+                self._reconnect_server()
+        if last_exc is not None:
+            raise last_exc
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self._safe_receive_world_update(retries=1)
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self._safe_receive_world_update(retries=2)
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+
+        orientation_quat_inv = R.from_quat(self.Player.robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(self.Player.robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.3
+        if is_fallen:
+            # remain = max(0, 800 - self.step_counter)
+            # return -8.0 - 0.01 * remain
+            return -1.0
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 2.0
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.01 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -0.3 * (tilt_mag)
+        ang_vel_penalty = -0.02 * ang_vel_mag
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -0.5 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 1200:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=40, kd=1.0
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = int(os.environ.get("GYM_CPU_N_ENVS", "20"))
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        server_warmup_sec = float(os.environ.get("GYM_CPU_SERVER_WARMUP_SEC", "3.0"))
+        n_steps_per_env = 256  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = 512  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = 1e-4
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env, monitor=False):
+            def thunk():
+                env = WalkEnv(self.ip, self.server_p + i_env)
+                if monitor:
+                    env = Monitor(env)
+                return env
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        if server_warmup_sec > 0:
+            print(f"Waiting {server_warmup_sec:.1f}s for server warmup...")
+            sleep(server_warmup_sec)
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i, monitor=True) for i in range(n_envs)])
+        # Use single-process eval env to avoid extra subprocess fragility during callback evaluation.
+        eval_env = DummyVecEnv([init_env(n_envs, monitor=True)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=0.03,  # Entropy coefficient for exploration
+                    clip_range=0.13,  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=0.95 , # Discount factor
+                    target_kl=0.03,
+                    n_epochs=5,
+                    tensorboard_log=f"./scripts/gyms/logs/{folder_name}/tensorboard/"
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 10, save_freq=n_steps_per_env * 10,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        server = Train_Server(self.server_p - 1, self.monitor_p, 1)
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+    trainer.train({"model_file": "scripts/gyms/logs/Walk_R0_004/best_model.zip"})
+    # trainer.test({"model_file": "scripts/gyms/logs/Walk_R0_004/best_model.zip",
+                #   "folder_dir": "scripts/gyms/logs/Walk_R0_004/",})

scripts/gyms/logs/Walk_R0_006/Walk.py

@@ -0,0 +1,679 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.monitor import Monitor
+from stable_baselines3.common.vec_env import SubprocVecEnv, DummyVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = True
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.015
+        self.reset_perturb_steps = 3
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def _reconnect_server(self):
+        try:
+            self.Player.server.shutdown()
+        except Exception:
+            pass
+
+        self.Player.server.connect()
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+
+    def _safe_receive_world_update(self, retries=1):
+        last_exc = None
+        for attempt in range(retries + 1):
+            try:
+                self.Player.server.receive()
+                self.Player.world.update()
+                return
+            except (ConnectionResetError, OSError) as exc:
+                last_exc = exc
+                if attempt >= retries:
+                    raise
+                self._reconnect_server()
+        if last_exc is not None:
+            raise last_exc
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self._safe_receive_world_update(retries=1)
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self._safe_receive_world_update(retries=2)
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+
+        orientation_quat_inv = R.from_quat(self.Player.robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(self.Player.robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.3
+        if is_fallen:
+            # remain = max(0, 800 - self.step_counter)
+            # return -8.0 - 0.01 * remain
+            return -1.0
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 2.0
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.01 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -0.3 * (tilt_mag)
+        ang_vel_penalty = -0.02 * ang_vel_mag
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -0.5 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 600:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=40, kd=1.0
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = int(os.environ.get("GYM_CPU_N_ENVS", "20"))
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        server_warmup_sec = float(os.environ.get("GYM_CPU_SERVER_WARMUP_SEC", "3.0"))
+        n_steps_per_env = int(os.environ.get("GYM_CPU_TRAIN_STEPS_PER_ENV", "256"))  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = int(os.environ.get("GYM_CPU_TRAIN_BATCH_SIZE", "512"))  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = float(os.environ.get("GYM_CPU_TRAIN_LR", "3e-4"))
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env, monitor=False):
+            def thunk():
+                env = WalkEnv(self.ip, self.server_p + i_env)
+                if monitor:
+                    env = Monitor(env)
+                return env
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1, no_render=True, no_realtime=True)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        if server_warmup_sec > 0:
+            print(f"Waiting {server_warmup_sec:.1f}s for server warmup...")
+            sleep(server_warmup_sec)
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i, monitor=True) for i in range(n_envs)])
+        # Use single-process eval env to avoid extra subprocess fragility during callback evaluation.
+        eval_env = DummyVecEnv([init_env(n_envs, monitor=True)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=float(os.environ.get("GYM_CPU_TRAIN_ENT_COEF", "0.05")),  # Entropy coefficient for exploration
+                    clip_range=float(os.environ.get("GYM_CPU_TRAIN_CLIP_RANGE", "0.2")),  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=float(os.environ.get("GYM_CPU_TRAIN_GAMMA", "0.95")), # Discount factor
+                    target_kl=0.03,
+                    n_epochs=int(os.environ.get("GYM_CPU_TRAIN_EPOCHS", "5")),
+                    # tensorboard_log=f"./scripts/gyms/logs/{folder_name}/tensorboard/"
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 20, save_freq=n_steps_per_env * 20,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        test_no_render = os.environ.get("GYM_CPU_TEST_NO_RENDER", "0") == "1"
+        test_no_realtime = os.environ.get("GYM_CPU_TEST_NO_REALTIME", "0") == "1"
+
+        server = Train_Server(
+            self.server_p - 1,
+            self.monitor_p,
+            1,
+            no_render=test_no_render,
+            no_realtime=test_no_realtime,
+        )
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+
+    run_mode = os.environ.get("GYM_CPU_MODE", "train").strip().lower()
+
+    if run_mode == "test":
+        test_model_file = os.environ.get("GYM_CPU_TEST_MODEL", "scripts/gyms/logs/Walk_R0_004/best_model.zip")
+        test_folder = os.environ.get("GYM_CPU_TEST_FOLDER", "scripts/gyms/logs/Walk_R0_004/")
+        trainer.test({"model_file": test_model_file, "folder_dir": test_folder})
+    else:
+        retrain_model = os.environ.get("GYM_CPU_TRAIN_MODEL", "").strip()
+        if retrain_model:
+            trainer.train({"model_file": retrain_model})
+        else:
+            trainer.train({})

scripts/gyms/logs/Walk_R0_007/Walk.py

@@ -0,0 +1,679 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.monitor import Monitor
+from stable_baselines3.common.vec_env import SubprocVecEnv, DummyVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = True
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.015
+        self.reset_perturb_steps = 3
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def _reconnect_server(self):
+        try:
+            self.Player.server.shutdown()
+        except Exception:
+            pass
+
+        self.Player.server.connect()
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+
+    def _safe_receive_world_update(self, retries=1):
+        last_exc = None
+        for attempt in range(retries + 1):
+            try:
+                self.Player.server.receive()
+                self.Player.world.update()
+                return
+            except (ConnectionResetError, OSError) as exc:
+                last_exc = exc
+                if attempt >= retries:
+                    raise
+                self._reconnect_server()
+        if last_exc is not None:
+            raise last_exc
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self._safe_receive_world_update(retries=1)
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self._safe_receive_world_update(retries=2)
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+
+        orientation_quat_inv = R.from_quat(self.Player.robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(self.Player.robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.3
+        if is_fallen:
+            # remain = max(0, 800 - self.step_counter)
+            # return -8.0 - 0.01 * remain
+            return -1.0
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 2.0
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.01 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -0.3 * (tilt_mag)
+        ang_vel_penalty = -0.02 * ang_vel_mag
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -1 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 1200:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=40, kd=1.0
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = int(os.environ.get("GYM_CPU_N_ENVS", "20"))
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        server_warmup_sec = float(os.environ.get("GYM_CPU_SERVER_WARMUP_SEC", "3.0"))
+        n_steps_per_env = int(os.environ.get("GYM_CPU_TRAIN_STEPS_PER_ENV", "256"))  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = int(os.environ.get("GYM_CPU_TRAIN_BATCH_SIZE", "512"))  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = float(os.environ.get("GYM_CPU_TRAIN_LR", "3e-4"))
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env, monitor=False):
+            def thunk():
+                env = WalkEnv(self.ip, self.server_p + i_env)
+                if monitor:
+                    env = Monitor(env)
+                return env
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1, no_render=True, no_realtime=True)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        if server_warmup_sec > 0:
+            print(f"Waiting {server_warmup_sec:.1f}s for server warmup...")
+            sleep(server_warmup_sec)
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i, monitor=True) for i in range(n_envs)])
+        # Use single-process eval env to avoid extra subprocess fragility during callback evaluation.
+        eval_env = DummyVecEnv([init_env(n_envs, monitor=True)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=float(os.environ.get("GYM_CPU_TRAIN_ENT_COEF", "0.05")),  # Entropy coefficient for exploration
+                    clip_range=float(os.environ.get("GYM_CPU_TRAIN_CLIP_RANGE", "0.2")),  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=float(os.environ.get("GYM_CPU_TRAIN_GAMMA", "0.95")), # Discount factor
+                    target_kl=0.03,
+                    n_epochs=int(os.environ.get("GYM_CPU_TRAIN_EPOCHS", "5")),
+                    # tensorboard_log=f"./scripts/gyms/logs/{folder_name}/tensorboard/"
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 20, save_freq=n_steps_per_env * 20, eval_eps=100,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        test_no_render = os.environ.get("GYM_CPU_TEST_NO_RENDER", "0") == "1"
+        test_no_realtime = os.environ.get("GYM_CPU_TEST_NO_REALTIME", "0") == "1"
+
+        server = Train_Server(
+            self.server_p - 1,
+            self.monitor_p,
+            1,
+            no_render=test_no_render,
+            no_realtime=test_no_realtime,
+        )
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+
+    run_mode = os.environ.get("GYM_CPU_MODE", "train").strip().lower()
+
+    if run_mode == "test":
+        test_model_file = os.environ.get("GYM_CPU_TEST_MODEL", "scripts/gyms/logs/Walk_R0_004/best_model.zip")
+        test_folder = os.environ.get("GYM_CPU_TEST_FOLDER", "scripts/gyms/logs/Walk_R0_004/")
+        trainer.test({"model_file": test_model_file, "folder_dir": test_folder})
+    else:
+        retrain_model = os.environ.get("GYM_CPU_TRAIN_MODEL", "").strip()
+        if retrain_model:
+            trainer.train({"model_file": retrain_model})
+        else:
+            trainer.train({})

scripts/gyms/logs/Walk_R0_008/Walk.py

@@ -0,0 +1,704 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.monitor import Monitor
+from stable_baselines3.common.vec_env import SubprocVecEnv, DummyVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Encourage a minimum lateral stance so the policy avoids feet overlap.
+        self.min_stance_rad = 0.10
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = False
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.015
+        self.reset_perturb_steps = 3
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def _reconnect_server(self):
+        try:
+            self.Player.server.shutdown()
+        except Exception:
+            pass
+
+        self.Player.server.connect()
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+
+    def _safe_receive_world_update(self, retries=1):
+        last_exc = None
+        for attempt in range(retries + 1):
+            try:
+                self.Player.server.receive()
+                self.Player.world.update()
+                return
+            except (ConnectionResetError, OSError) as exc:
+                last_exc = exc
+                if attempt >= retries:
+                    raise
+                self._reconnect_server()
+        if last_exc is not None:
+            raise last_exc
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self._safe_receive_world_update(retries=1)
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self._safe_receive_world_update(retries=2)
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+        robot = self.Player.robot
+
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.3
+        if is_fallen:
+            # remain = max(0, 800 - self.step_counter)
+            # return -8.0 - 0.01 * remain
+            return -1.0
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 2.0
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.01 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -0.3 * (tilt_mag)
+        ang_vel_penalty = -0.02 * ang_vel_mag
+
+        # Use simulator joint readings in training frame to shape lateral stance.
+        joint_pos = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        ) * self.train_sim_flip
+        left_hip_roll = float(joint_pos[12])
+        right_hip_roll = float(joint_pos[18])
+        left_ankle_roll = float(joint_pos[16])
+        right_ankle_roll = float(joint_pos[22])
+
+        hip_spread = left_hip_roll - right_hip_roll
+        ankle_spread = left_ankle_roll - right_ankle_roll
+        stance_metric = 0.6 * abs(hip_spread) + 0.4 * abs(ankle_spread)
+
+        # Penalize narrow stance (feet too close) and scissoring (cross-leg pattern).
+        stance_collapse_penalty = -4.0 * max(0.0, self.min_stance_rad - stance_metric)
+        cross_leg_penalty = -1.2 * max(0.0, -(hip_spread * ankle_spread))
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -0.5 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                 + stance_collapse_penalty
+                 + cross_leg_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 600:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                  f"stance_collapse_penalty:{stance_collapse_penalty:.4f}",
+                  f"cross_leg_penalty:{cross_leg_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=40, kd=1.0
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = int(os.environ.get("GYM_CPU_N_ENVS", "20"))
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        server_warmup_sec = float(os.environ.get("GYM_CPU_SERVER_WARMUP_SEC", "3.0"))
+        n_steps_per_env = int(os.environ.get("GYM_CPU_TRAIN_STEPS_PER_ENV", "256"))  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = int(os.environ.get("GYM_CPU_TRAIN_BATCH_SIZE", "512"))  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = float(os.environ.get("GYM_CPU_TRAIN_LR", "3e-4"))
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env, monitor=False):
+            def thunk():
+                env = WalkEnv(self.ip, self.server_p + i_env)
+                if monitor:
+                    env = Monitor(env)
+                return env
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1, no_render=True, no_realtime=True)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        if server_warmup_sec > 0:
+            print(f"Waiting {server_warmup_sec:.1f}s for server warmup...")
+            sleep(server_warmup_sec)
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i, monitor=True) for i in range(n_envs)])
+        # Use single-process eval env to avoid extra subprocess fragility during callback evaluation.
+        eval_env = DummyVecEnv([init_env(n_envs, monitor=True)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=float(os.environ.get("GYM_CPU_TRAIN_ENT_COEF", "0.05")),  # Entropy coefficient for exploration
+                    clip_range=float(os.environ.get("GYM_CPU_TRAIN_CLIP_RANGE", "0.2")),  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=float(os.environ.get("GYM_CPU_TRAIN_GAMMA", "0.95")), # Discount factor
+                    target_kl=0.03,
+                    n_epochs=int(os.environ.get("GYM_CPU_TRAIN_EPOCHS", "5")),
+                    tensorboard_log=f"./scripts/gyms/logs/{folder_name}/tensorboard/"
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 20, save_freq=n_steps_per_env * 20, eval_eps=100,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        test_no_render = os.environ.get("GYM_CPU_TEST_NO_RENDER", "0") == "1"
+        test_no_realtime = os.environ.get("GYM_CPU_TEST_NO_REALTIME", "0") == "1"
+
+        server = Train_Server(
+            self.server_p - 1,
+            self.monitor_p,
+            1,
+            no_render=test_no_render,
+            no_realtime=test_no_realtime,
+        )
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+
+    run_mode = os.environ.get("GYM_CPU_MODE", "train").strip().lower()
+
+    if run_mode == "test":
+        test_model_file = os.environ.get("GYM_CPU_TEST_MODEL", "scripts/gyms/logs/Walk_R0_004/best_model.zip")
+        test_folder = os.environ.get("GYM_CPU_TEST_FOLDER", "scripts/gyms/logs/Walk_R0_004/")
+        trainer.test({"model_file": test_model_file, "folder_dir": test_folder})
+    else:
+        retrain_model = os.environ.get("GYM_CPU_TRAIN_MODEL", "").strip()
+        if retrain_model:
+            trainer.train({"model_file": retrain_model})
+        else:
+            trainer.train({})

scripts/gyms/logs/Walk_R0_009/Walk.py

@@ -0,0 +1,705 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+from itertools import count
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.monitor import Monitor
+from stable_baselines3.common.vec_env import SubprocVecEnv, DummyVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Encourage a minimum lateral stance so the policy avoids feet overlap.
+        self.min_stance_rad = 0.10
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = False
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.015
+        self.reset_perturb_steps = 3
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def _reconnect_server(self):
+        try:
+            self.Player.server.shutdown()
+        except Exception:
+            pass
+
+        self.Player.server.connect()
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+
+    def _safe_receive_world_update(self, retries=1):
+        last_exc = None
+        for attempt in range(retries + 1):
+            try:
+                self.Player.server.receive()
+                self.Player.world.update()
+                return
+            except (ConnectionResetError, OSError) as exc:
+                last_exc = exc
+                if attempt >= retries:
+                    raise
+                self._reconnect_server()
+        if last_exc is not None:
+            raise last_exc
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self._safe_receive_world_update(retries=1)
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self._safe_receive_world_update(retries=2)
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+        robot = self.Player.robot
+
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.3
+        if is_fallen:
+            # remain = max(0, 800 - self.step_counter)
+            # return -8.0 - 0.01 * remain
+            return -1.0
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 2.0
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.01 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -0.3 * (tilt_mag)
+        ang_vel_penalty = -0.02 * ang_vel_mag
+
+        # Use simulator joint readings in training frame to shape lateral stance.
+        joint_pos = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        ) * self.train_sim_flip
+        left_hip_roll = float(joint_pos[12])
+        right_hip_roll = float(joint_pos[18])
+        left_ankle_roll = float(joint_pos[16])
+        right_ankle_roll = float(joint_pos[22])
+
+        hip_spread = left_hip_roll - right_hip_roll
+        ankle_spread = left_ankle_roll - right_ankle_roll
+        stance_metric = 0.6 * abs(hip_spread) + 0.4 * abs(ankle_spread)
+
+        # Penalize narrow stance (feet too close) and scissoring (cross-leg pattern).
+        stance_collapse_penalty = -4.0 * max(0.0, self.min_stance_rad - stance_metric)
+        cross_leg_penalty = -1.2 * max(0.0, -(hip_spread * ankle_spread))
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -0.5 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                 + stance_collapse_penalty
+                 + cross_leg_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 600:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                  f"stance_collapse_penalty:{stance_collapse_penalty:.4f}",
+                  f"cross_leg_penalty:{cross_leg_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=40, kd=1.0
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = int(os.environ.get("GYM_CPU_N_ENVS", "20"))
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        server_warmup_sec = float(os.environ.get("GYM_CPU_SERVER_WARMUP_SEC", "3.0"))
+        n_steps_per_env = int(os.environ.get("GYM_CPU_TRAIN_STEPS_PER_ENV", "256"))  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = int(os.environ.get("GYM_CPU_TRAIN_BATCH_SIZE", "512"))  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = float(os.environ.get("GYM_CPU_TRAIN_LR", "3e-4"))
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env, monitor=False):
+            def thunk():
+                env = WalkEnv(self.ip, self.server_p + i_env)
+                if monitor:
+                    env = Monitor(env)
+                return env
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1, no_render=True, no_realtime=True)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        if server_warmup_sec > 0:
+            print(f"Waiting {server_warmup_sec:.1f}s for server warmup...")
+            sleep(server_warmup_sec)
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i, monitor=True) for i in range(n_envs)])
+        # Use single-process eval env to avoid extra subprocess fragility during callback evaluation.
+        eval_env = DummyVecEnv([init_env(n_envs, monitor=True)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=float(os.environ.get("GYM_CPU_TRAIN_ENT_COEF", "0.05")),  # Entropy coefficient for exploration
+                    clip_range=float(os.environ.get("GYM_CPU_TRAIN_CLIP_RANGE", "0.2")),  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=float(os.environ.get("GYM_CPU_TRAIN_GAMMA", "0.95")), # Discount factor
+                    # target_kl=0.03,
+                    n_epochs=int(os.environ.get("GYM_CPU_TRAIN_EPOCHS", "5")),
+                    tensorboard_log=f"./scripts/gyms/logs/{folder_name}/tensorboard/"
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 20, save_freq=n_steps_per_env * 20, eval_eps=100,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        test_no_render = os.environ.get("GYM_CPU_TEST_NO_RENDER", "0") == "1"
+        test_no_realtime = os.environ.get("GYM_CPU_TEST_NO_REALTIME", "0") == "1"
+
+        server = Train_Server(
+            self.server_p - 1,
+            self.monitor_p,
+            1,
+            no_render=test_no_render,
+            no_realtime=test_no_realtime,
+        )
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+
+    run_mode = os.environ.get("GYM_CPU_MODE", "train").strip().lower()
+
+    if run_mode == "test":
+        test_model_file = os.environ.get("GYM_CPU_TEST_MODEL", "scripts/gyms/logs/Walk_R0_004/best_model.zip")
+        test_folder = os.environ.get("GYM_CPU_TEST_FOLDER", "scripts/gyms/logs/Walk_R0_004/")
+        trainer.test({"model_file": test_model_file, "folder_dir": test_folder})
+    else:
+        retrain_model = os.environ.get("GYM_CPU_TRAIN_MODEL", "").strip()
+        if retrain_model:
+            trainer.train({"model_file": retrain_model})
+        else:
+            trainer.train({})

scripts/gyms/logs/Walk_R0_010/Walk.py

@@ -0,0 +1,705 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+from itertools import count
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.monitor import Monitor
+from stable_baselines3.common.vec_env import SubprocVecEnv, DummyVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Encourage a minimum lateral stance so the policy avoids feet overlap.
+        self.min_stance_rad = 0.10
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = False
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.015
+        self.reset_perturb_steps = 3
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def _reconnect_server(self):
+        try:
+            self.Player.server.shutdown()
+        except Exception:
+            pass
+
+        self.Player.server.connect()
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+
+    def _safe_receive_world_update(self, retries=1):
+        last_exc = None
+        for attempt in range(retries + 1):
+            try:
+                self.Player.server.receive()
+                self.Player.world.update()
+                return
+            except (ConnectionResetError, OSError) as exc:
+                last_exc = exc
+                if attempt >= retries:
+                    raise
+                self._reconnect_server()
+        if last_exc is not None:
+            raise last_exc
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self._safe_receive_world_update(retries=1)
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self._safe_receive_world_update(retries=2)
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+        robot = self.Player.robot
+
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.55
+        if is_fallen:
+            # remain = max(0, 800 - self.step_counter)
+            # return -8.0 - 0.01 * remain
+            return -1.0
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 2.0
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.01 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -0.3 * (tilt_mag)
+        ang_vel_penalty = -0.02 * ang_vel_mag
+
+        # Use simulator joint readings in training frame to shape lateral stance.
+        joint_pos = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        ) * self.train_sim_flip
+        left_hip_roll = float(joint_pos[12])
+        right_hip_roll = float(joint_pos[18])
+        left_ankle_roll = float(joint_pos[16])
+        right_ankle_roll = float(joint_pos[22])
+
+        hip_spread = left_hip_roll - right_hip_roll
+        ankle_spread = left_ankle_roll - right_ankle_roll
+        stance_metric = 0.6 * abs(hip_spread) + 0.4 * abs(ankle_spread)
+
+        # Penalize narrow stance (feet too close) and scissoring (cross-leg pattern).
+        stance_collapse_penalty = -4.0 * max(0.0, self.min_stance_rad - stance_metric)
+        cross_leg_penalty = -1.2 * max(0.0, -(hip_spread * ankle_spread))
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -0.5 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                 + stance_collapse_penalty
+                 + cross_leg_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 600:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                  f"stance_collapse_penalty:{stance_collapse_penalty:.4f}",
+                  f"cross_leg_penalty:{cross_leg_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=40, kd=1.0
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.55
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = int(os.environ.get("GYM_CPU_N_ENVS", "20"))
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        server_warmup_sec = float(os.environ.get("GYM_CPU_SERVER_WARMUP_SEC", "3.0"))
+        n_steps_per_env = int(os.environ.get("GYM_CPU_TRAIN_STEPS_PER_ENV", "256"))  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = int(os.environ.get("GYM_CPU_TRAIN_BATCH_SIZE", "512"))  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = float(os.environ.get("GYM_CPU_TRAIN_LR", "3e-4"))
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env, monitor=False):
+            def thunk():
+                env = WalkEnv(self.ip, self.server_p + i_env)
+                if monitor:
+                    env = Monitor(env)
+                return env
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1, no_render=True, no_realtime=True)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        if server_warmup_sec > 0:
+            print(f"Waiting {server_warmup_sec:.1f}s for server warmup...")
+            sleep(server_warmup_sec)
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i, monitor=True) for i in range(n_envs)])
+        # Use single-process eval env to avoid extra subprocess fragility during callback evaluation.
+        eval_env = DummyVecEnv([init_env(n_envs, monitor=True)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=float(os.environ.get("GYM_CPU_TRAIN_ENT_COEF", "0.05")),  # Entropy coefficient for exploration
+                    clip_range=float(os.environ.get("GYM_CPU_TRAIN_CLIP_RANGE", "0.2")),  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=float(os.environ.get("GYM_CPU_TRAIN_GAMMA", "0.95")), # Discount factor
+                    # target_kl=0.03,
+                    n_epochs=int(os.environ.get("GYM_CPU_TRAIN_EPOCHS", "5")),
+                    tensorboard_log=f"./scripts/gyms/logs/{folder_name}/tensorboard/"
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 20, save_freq=n_steps_per_env * 20, eval_eps=100,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        test_no_render = os.environ.get("GYM_CPU_TEST_NO_RENDER", "0") == "1"
+        test_no_realtime = os.environ.get("GYM_CPU_TEST_NO_REALTIME", "0") == "1"
+
+        server = Train_Server(
+            self.server_p - 1,
+            self.monitor_p,
+            1,
+            no_render=test_no_render,
+            no_realtime=test_no_realtime,
+        )
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+
+    run_mode = os.environ.get("GYM_CPU_MODE", "train").strip().lower()
+
+    if run_mode == "test":
+        test_model_file = os.environ.get("GYM_CPU_TEST_MODEL", "scripts/gyms/logs/Walk_R0_004/best_model.zip")
+        test_folder = os.environ.get("GYM_CPU_TEST_FOLDER", "scripts/gyms/logs/Walk_R0_004/")
+        trainer.test({"model_file": test_model_file, "folder_dir": test_folder})
+    else:
+        retrain_model = os.environ.get("GYM_CPU_TRAIN_MODEL", "").strip()
+        if retrain_model:
+            trainer.train({"model_file": retrain_model})
+        else:
+            trainer.train({})

scripts/gyms/logs/stand_stable_0.1/Walk.py

@@ -0,0 +1,626 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.vec_env import SubprocVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = True
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.015
+        self.reset_perturb_steps = 3
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self.Player.server.receive()
+        self.Player.world.update()
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self.Player.server.receive()
+            self.Player.world.update()
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+
+        orientation_quat_inv = R.from_quat(self.Player.robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(self.Player.robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.3
+        if is_fallen:
+            # remain = max(0, 800 - self.step_counter)
+            # return -8.0 - 0.01 * remain
+            return -1.0
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 2.0
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.01 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -0.3 * (tilt_mag)
+        ang_vel_penalty = -0.02 * ang_vel_mag
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -0.5 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 1200:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=40, kd=1.0
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = 20  # Reduced from 8 to decrease CPU/network pressure during init
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        n_steps_per_env = 256  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = 512  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = 3e-4
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env):
+            def thunk():
+                return WalkEnv(self.ip, self.server_p + i_env)
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i) for i in range(n_envs)])
+        eval_env = SubprocVecEnv([init_env(n_envs)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=0.05,  # Entropy coefficient for exploration
+                    # clip_range=0.13,  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=0.95 , # Discount factor
+                    target_kl=0.03,
+                    n_epochs=5
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 10, save_freq=n_steps_per_env * 10,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        server = Train_Server(self.server_p - 1, self.monitor_p, 1)
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+    trainer.train({})
+    # trainer.test({"model_file": "scripts/gyms/logs/Walk_R0_000/best_model.zip",
+                #   "folder_dir": "scripts/gyms/logs/Walk_R0_000/",})

scripts/gyms/logs/stand_stable_final/Walk.py

@@ -0,0 +1,705 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+from itertools import count
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.monitor import Monitor
+from stable_baselines3.common.vec_env import SubprocVecEnv, DummyVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Encourage a minimum lateral stance so the policy avoids feet overlap.
+        self.min_stance_rad = 0.10
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = False
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.015
+        self.reset_perturb_steps = 3
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def _reconnect_server(self):
+        try:
+            self.Player.server.shutdown()
+        except Exception:
+            pass
+
+        self.Player.server.connect()
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+
+    def _safe_receive_world_update(self, retries=1):
+        last_exc = None
+        for attempt in range(retries + 1):
+            try:
+                self.Player.server.receive()
+                self.Player.world.update()
+                return
+            except (ConnectionResetError, OSError) as exc:
+                last_exc = exc
+                if attempt >= retries:
+                    raise
+                self._reconnect_server()
+        if last_exc is not None:
+            raise last_exc
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self._safe_receive_world_update(retries=1)
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self._safe_receive_world_update(retries=2)
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+        robot = self.Player.robot
+
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.55
+        if is_fallen:
+            # remain = max(0, 800 - self.step_counter)
+            # return -8.0 - 0.01 * remain
+            return -1.0
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 2.0
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.01 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -0.3 * (tilt_mag)
+        ang_vel_penalty = -0.02 * ang_vel_mag
+
+        # Use simulator joint readings in training frame to shape lateral stance.
+        joint_pos = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        ) * self.train_sim_flip
+        left_hip_roll = float(joint_pos[12])
+        right_hip_roll = float(joint_pos[18])
+        left_ankle_roll = float(joint_pos[16])
+        right_ankle_roll = float(joint_pos[22])
+
+        hip_spread = left_hip_roll - right_hip_roll
+        ankle_spread = left_ankle_roll - right_ankle_roll
+        stance_metric = 0.6 * abs(hip_spread) + 0.4 * abs(ankle_spread)
+
+        # Penalize narrow stance (feet too close) and scissoring (cross-leg pattern).
+        stance_collapse_penalty = -4.0 * max(0.0, self.min_stance_rad - stance_metric)
+        cross_leg_penalty = -1.2 * max(0.0, -(hip_spread * ankle_spread))
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -0.5 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                 + stance_collapse_penalty
+                 + cross_leg_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 600:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                  f"stance_collapse_penalty:{stance_collapse_penalty:.4f}",
+                  f"cross_leg_penalty:{cross_leg_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=40, kd=1.0
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.55
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = int(os.environ.get("GYM_CPU_N_ENVS", "20"))
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        server_warmup_sec = float(os.environ.get("GYM_CPU_SERVER_WARMUP_SEC", "3.0"))
+        n_steps_per_env = int(os.environ.get("GYM_CPU_TRAIN_STEPS_PER_ENV", "256"))  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = int(os.environ.get("GYM_CPU_TRAIN_BATCH_SIZE", "512"))  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = float(os.environ.get("GYM_CPU_TRAIN_LR", "3e-4"))
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env, monitor=False):
+            def thunk():
+                env = WalkEnv(self.ip, self.server_p + i_env)
+                if monitor:
+                    env = Monitor(env)
+                return env
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1, no_render=True, no_realtime=True)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        if server_warmup_sec > 0:
+            print(f"Waiting {server_warmup_sec:.1f}s for server warmup...")
+            sleep(server_warmup_sec)
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i, monitor=True) for i in range(n_envs)])
+        # Use single-process eval env to avoid extra subprocess fragility during callback evaluation.
+        eval_env = DummyVecEnv([init_env(n_envs, monitor=True)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=float(os.environ.get("GYM_CPU_TRAIN_ENT_COEF", "0.05")),  # Entropy coefficient for exploration
+                    clip_range=float(os.environ.get("GYM_CPU_TRAIN_CLIP_RANGE", "0.2")),  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=float(os.environ.get("GYM_CPU_TRAIN_GAMMA", "0.95")), # Discount factor
+                    # target_kl=0.03,
+                    n_epochs=int(os.environ.get("GYM_CPU_TRAIN_EPOCHS", "5")),
+                    tensorboard_log=f"./scripts/gyms/logs/{folder_name}/tensorboard/"
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 20, save_freq=n_steps_per_env * 20, eval_eps=100,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        test_no_render = os.environ.get("GYM_CPU_TEST_NO_RENDER", "0") == "1"
+        test_no_realtime = os.environ.get("GYM_CPU_TEST_NO_REALTIME", "0") == "1"
+
+        server = Train_Server(
+            self.server_p - 1,
+            self.monitor_p,
+            1,
+            no_render=test_no_render,
+            no_realtime=test_no_realtime,
+        )
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+
+    run_mode = os.environ.get("GYM_CPU_MODE", "train").strip().lower()
+
+    if run_mode == "test":
+        test_model_file = os.environ.get("GYM_CPU_TEST_MODEL", "scripts/gyms/logs/Walk_R0_004/best_model.zip")
+        test_folder = os.environ.get("GYM_CPU_TEST_FOLDER", "scripts/gyms/logs/Walk_R0_004/")
+        trainer.test({"model_file": test_model_file, "folder_dir": test_folder})
+    else:
+        retrain_model = os.environ.get("GYM_CPU_TRAIN_MODEL", "").strip()
+        if retrain_model:
+            trainer.train({"model_file": retrain_model})
+        else:
+            trainer.train({})

train.sh

@@ -0,0 +1,136 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+# ------------------------------
+# 资源限制配置（cgroup v2 + systemd-run）
+# ------------------------------
+# 说明：
+# 1) 这个脚本会把训练进程放进一个临时的 systemd scope 中，并施加 CPU/内存上限。
+# 2) 仅限制“本次训练进程”，不会永久改系统配置。
+# 3) 下面变量都支持“环境变量覆盖”，即你可以在命令前临时指定。
+#
+# CPU 核数基准（默认 20）：
+# 例如你的机器按 20 核预算来算，可保持默认。
+CORES="${CORES:-20}"
+# CPU 占用百分比（默认 100）：
+# 最终会与 CORES 相乘得到 CPUQuota。
+# 例：CORES=20, UTIL_PERCENT=100 -> CPUQuota=2000%（约 20 核等效）
+UTIL_PERCENT="${UTIL_PERCENT:-100}"
+CPU_QUOTA="$((CORES * UTIL_PERCENT))%"
+
+# 内存上限（默认关闭）：
+# 设为具体值（如 24G/28G）可限制训练最多占用内存；
+# 设为 0/none/off/infinity 表示不设置 cgroup 内存上限。
+MEMORY_MAX="${MEMORY_MAX:-0}"
+
+# ------------------------------
+# 训练运行参数（由 scripts/gyms/Walk.py 读取）
+# ------------------------------
+# 运行模式：train 或 test
+GYM_CPU_MODE="${GYM_CPU_MODE:-train}"
+
+# 并行环境数量：越大通常吞吐越高，但也更容易触发 OOM 或连接不稳定。
+# 默认使用更稳妥的 12，确认稳定后再升到 16/20。
+GYM_CPU_N_ENVS="${GYM_CPU_N_ENVS:-20}"
+# 服务器预热时间（秒）：
+# 在批量拉起 rcssserver 后等待一段时间，再创建 SubprocVecEnv，
+# 可降低 ConnectionReset/EOFError 概率。
+GYM_CPU_SERVER_WARMUP_SEC="${GYM_CPU_SERVER_WARMUP_SEC:-10}"
+
+# 训练专用参数
+GYM_CPU_TRAIN_STEPS_PER_ENV="${GYM_CPU_TRAIN_STEPS_PER_ENV:-256}"
+GYM_CPU_TRAIN_BATCH_SIZE="${GYM_CPU_TRAIN_BATCH_SIZE:-512}"
+GYM_CPU_TRAIN_LR="${GYM_CPU_TRAIN_LR:-1e-4}"
+GYM_CPU_TRAIN_ENT_COEF="${GYM_CPU_TRAIN_ENT_COEF:-0.03}"
+GYM_CPU_TRAIN_CLIP_RANGE="${GYM_CPU_TRAIN_CLIP_RANGE:-0.13}"
+GYM_CPU_TRAIN_GAMMA="${GYM_CPU_TRAIN_GAMMA:-0.95}"
+GYM_CPU_TRAIN_EPOCHS="${GYM_CPU_TRAIN_EPOCHS:-5}"
+GYM_CPU_TRAIN_MODEL="${GYM_CPU_TRAIN_MODEL:-}"
+
+# 测试专用参数
+GYM_CPU_TEST_MODEL="${GYM_CPU_TEST_MODEL:-scripts/gyms/logs/Walk_R0_004/best_model.zip}"
+GYM_CPU_TEST_FOLDER="${GYM_CPU_TEST_FOLDER:-scripts/gyms/logs/Walk_R0_004/}"
+# 测试默认实时且显示画面：默认均为 0
+# 设为 1 表示关闭对应能力
+GYM_CPU_TEST_NO_RENDER="${GYM_CPU_TEST_NO_RENDER:-0}"
+GYM_CPU_TEST_NO_REALTIME="${GYM_CPU_TEST_NO_REALTIME:-0}"
+
+# Python 解释器选择策略：
+# 1) 优先使用你手动传入的 PYTHON_BIN
+# 2) 其次用当前激活 conda 环境（CONDA_PREFIX/bin/python）
+# 3) 再回退到默认 mujoco 环境路径
+# 4) 最后尝试系统 python / python3
+DEFAULT_PYTHON="/home/solren/Downloads/Anaconda/envs/mujoco/bin/python"
+CONDA_PYTHON="${CONDA_PREFIX:-}/bin/python"
+
+# 安全保护：不要用 sudo 运行。
+# 原因：sudo 可能导致 conda 环境与用户会话环境不一致，
+# 会引发 python 路径丢失、systemd --user 会话不可见等问题。
+if [[ "${EUID}" -eq 0 ]]; then
+	echo "Do not run this script with sudo; run as your normal user in conda env 'mujoco'."
+	exit 1
+fi
+
+# 解析最终使用的 Python 可执行文件。
+if [[ -n "${PYTHON_BIN:-}" ]]; then
+	PYTHON_EXEC="${PYTHON_BIN}"
+elif [[ -n "${CONDA_PREFIX:-}" && -x "${CONDA_PYTHON}" ]]; then
+	PYTHON_EXEC="${CONDA_PYTHON}"
+elif [[ -x "${DEFAULT_PYTHON}" ]]; then
+	PYTHON_EXEC="${DEFAULT_PYTHON}"
+elif command -v python >/dev/null 2>&1; then
+	PYTHON_EXEC="$(command -v python)"
+elif command -v python3 >/dev/null 2>&1; then
+	PYTHON_EXEC="$(command -v python3)"
+else
+	echo "No Python executable found. Set PYTHON_BIN=/abs/path/to/python and retry."
+	exit 1
+fi
+
+# 脚本所在目录（绝对路径），便于后续定位模块/相对路径。
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+
+# 打印当前生效配置，方便排障和复现实验。
+echo "Starting training with limits: CPU=${CPU_QUOTA}, Memory=${MEMORY_MAX}"
+echo "Mode: ${GYM_CPU_MODE}"
+echo "Runtime knobs: GYM_CPU_N_ENVS=${GYM_CPU_N_ENVS}, GYM_CPU_SERVER_WARMUP_SEC=${GYM_CPU_SERVER_WARMUP_SEC}"
+echo "Using Python: ${PYTHON_EXEC}"
+if [[ -n "${CONDA_DEFAULT_ENV:-}" ]]; then
+	echo "Detected conda env: ${CONDA_DEFAULT_ENV}"
+fi
+
+SYSTEMD_PROPS=("-p" "CPUQuota=${CPU_QUOTA}")
+case "${MEMORY_MAX,,}" in
+	0|none|off|infinity)
+		echo "MemoryMax is disabled for this run (no cgroup memory cap)."
+		;;
+	*)
+		SYSTEMD_PROPS+=("-p" "MemoryMax=${MEMORY_MAX}")
+		;;
+esac
+
+# 使用 systemd-run --user --scope 启动“受限资源”的训练进程：
+# - CPUQuota: 总 CPU 配额
+# - MemoryMax: 最大内存
+# - env ...  : 显式传递训练参数到 Python 进程
+# - python -m scripts.gyms.Walk: 以模块方式启动训练入口
+systemd-run --user --scope \
+	"${SYSTEMD_PROPS[@]}" \
+	env \
+	GYM_CPU_MODE="${GYM_CPU_MODE}" \
+	GYM_CPU_N_ENVS="${GYM_CPU_N_ENVS}" \
+	GYM_CPU_SERVER_WARMUP_SEC="${GYM_CPU_SERVER_WARMUP_SEC}" \
+	GYM_CPU_TRAIN_STEPS_PER_ENV="${GYM_CPU_TRAIN_STEPS_PER_ENV}" \
+	GYM_CPU_TRAIN_BATCH_SIZE="${GYM_CPU_TRAIN_BATCH_SIZE}" \
+	GYM_CPU_TRAIN_LR="${GYM_CPU_TRAIN_LR}" \
+	GYM_CPU_TRAIN_ENT_COEF="${GYM_CPU_TRAIN_ENT_COEF}" \
+	GYM_CPU_TRAIN_CLIP_RANGE="${GYM_CPU_TRAIN_CLIP_RANGE}" \
+	GYM_CPU_TRAIN_GAMMA="${GYM_CPU_TRAIN_GAMMA}" \
+	GYM_CPU_TRAIN_EPOCHS="${GYM_CPU_TRAIN_EPOCHS}" \
+	GYM_CPU_TRAIN_MODEL="${GYM_CPU_TRAIN_MODEL}" \
+	GYM_CPU_TEST_MODEL="${GYM_CPU_TEST_MODEL}" \
+	GYM_CPU_TEST_FOLDER="${GYM_CPU_TEST_FOLDER}" \
+	GYM_CPU_TEST_NO_RENDER="${GYM_CPU_TEST_NO_RENDER}" \
+	GYM_CPU_TEST_NO_REALTIME="${GYM_CPU_TEST_NO_REALTIME}" \
+	"${PYTHON_EXEC}" "-m" "scripts.gyms.Walk"
+

world/world.py

@@ -47,6 +47,7 @@ class World:
         self.their_team_players: list[OtherRobot] = [OtherRobot(is_teammate=False) for _ in
                                                      range(self.MAX_PLAYERS_PER_TEAM)]
         self.field: Field = self.__initialize_field(field_name=field_name)
+        self.WORLD_STEPTIME: float = 0.005  # Time step of the world in seconds
 
     def update(self) -> None:
         """