加入dqn算法

DQN/RL_brain.py

@@ -1,254 +0,0 @@
-"""
-Deep Q Network off-policy
-"""
-import torch
-import torch.nn as nn
-import numpy as np
-import pandas as pd
-import matplotlib.pyplot as plt
-
-np.random.seed(42)
-torch.manual_seed(2)
-
-
-class Network(nn.Module):
-    """
-    Network Structure
-    """
-    def __init__(self,
-                 n_features,
-                 n_actions,
-                 n_neuron=10
-                 ):
-        super(Network, self).__init__()
-        self.net = nn.Sequential(
-            nn.Linear(in_features=n_features, out_features=n_neuron, bias=True),
-            nn.Linear(in_features=n_neuron, out_features=n_actions, bias=True),
-            nn.ReLU()
-        )
-
-    def forward(self, s):
-        """
-
-        :param s: s
-        :return: q
-        """
-        q = self.net(s)
-        return q
-
-
-class DeepQNetwork(nn.Module):
-    """
-    Q Learning Algorithm
-    """
-    def __init__(self,
-                 n_actions,
-                 n_features,
-                 learning_rate=0.01,
-                 reward_decay=0.9,
-                 e_greedy=0.9,
-                 replace_target_iter=300,
-                 memory_size=500,
-                 batch_size=32,
-                 e_greedy_increment=None):
-        super(DeepQNetwork, self).__init__()
-
-        self.n_actions = n_actions
-        self.n_features = n_features
-        self.lr = learning_rate
-        self.gamma = reward_decay
-        self.epsilon_max = e_greedy
-        self.replace_target_iter = replace_target_iter
-        self.memory_size = memory_size
-        self.batch_size = batch_size
-        self.epsilon_increment = e_greedy_increment
-        self.epsilon = 0 if e_greedy_increment is not None else self.epsilon_max
-
-        # total learning step
-        self.learn_step_counter = 0
-
-        # initialize zero memory [s, a, r, s_]
-        # 这里用pd.DataFrame创建的表格作为memory
-        # 表格的行数是memory的大小，也就是transition的个数
-        # 表格的列数是transition的长度，一个transition包含[s, a, r, s_]，其中a和r分别是一个数字，s和s_的长度分别是n_features
-        self.memory = pd.DataFrame(np.zeros((self.memory_size, self.n_features*2+2)))
-
-        # build two network: eval_net and target_net
-        self.eval_net = Network(n_features=self.n_features, n_actions=self.n_actions)
-        self.target_net = Network(n_features=self.n_features, n_actions=self.n_actions)
-        self.loss_function = nn.MSELoss()
-        self.optimizer = torch.optim.Adam(self.eval_net.parameters(), lr=self.lr)
-
-        # 记录每一步的误差
-        self.cost_his = []
-
-
-    def store_transition(self, s, a, r, s_):
-        if not hasattr(self, 'memory_counter'):
-            # hasattr用于判断对象是否包含对应的属性。
-            self.memory_counter = 0
-
-        transition = np.hstack((s, [a,r], s_))
-
-        # replace the old memory with new memory
-        index = self.memory_counter % self.memory_size
-        self.memory.iloc[index, :] = transition
-
-        self.memory_counter += 1
-
-    def choose_action(self, observation):
-        observation = observation[np.newaxis, :]
-
-        if np.random.uniform() < self.epsilon:
-            # forward feed the observation and get q value for every actions
-            s = torch.FloatTensor(observation)
-            actions_value = self.eval_net(s)
-            action = [np.argmax(actions_value.detach().numpy())][0]
-        else:
-            action = np.random.randint(0, self.n_actions)
-        return action
-
-    def _replace_target_params(self):
-        # 复制网络参数
-        self.target_net.load_state_dict(self.eval_net.state_dict())
-
-    def learn(self):
-        # check to replace target parameters
-        if self.learn_step_counter % self.replace_target_iter == 0:
-            self._replace_target_params()
-            print('\ntarget params replaced\n')
-
-        # sample batch memory from all memory
-        batch_memory = self.memory.sample(self.batch_size) \
-            if self.memory_counter > self.memory_size \
-            else self.memory.iloc[:self.memory_counter].sample(self.batch_size, replace=True)
-
-        # run the nextwork
-        s = torch.FloatTensor(batch_memory.iloc[:, :self.n_features].values)
-        s_ = torch.FloatTensor(batch_memory.iloc[:, -self.n_features:].values)
-        q_eval = self.eval_net(s)
-        q_next = self.target_net(s_)
-
-        # change q_target w.r.t q_eval's action
-        q_target = q_eval.clone()
-
-        # 更新值
-        batch_index = np.arange(self.batch_size, dtype=np.int32)
-        eval_act_index = batch_memory.iloc[:, self.n_features].values.astype(int)
-        reward = batch_memory.iloc[:, self.n_features + 1].values
-
-        q_target[batch_index, eval_act_index] = torch.FloatTensor(reward) + self.gamma * q_next.max(dim=1).values
-
-        # train eval network
-        loss = self.loss_function(q_target, q_eval)
-        self.optimizer.zero_grad()
-        loss.backward()
-        self.optimizer.step()
-
-        self.cost_his.append(loss.detach().numpy())
-
-        # increasing epsilon
-        self.epsilon = self.epsilon + self.epsilon_increment if self.epsilon < self.epsilon_max else self.epsilon_max
-        self.learn_step_counter += 1
-
-    def plot_cost(self):
-        plt.figure()
-        plt.plot(np.arange(len(self.cost_his)), self.cost_his)
-        plt.show()
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-

DQN/run_this.py

@@ -1,58 +0,0 @@
-from RL_brain import DeepQNetwork
-import os
-import sys
-sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
-from env import PartitionMazeEnv
-
-def run_maze():
-    step = 0  # 为了记录走到第几步，记忆录中积累经验（也就是积累一些transition）之后再开始学习
-    for episode in range(200):
-        # initial observation
-        observation = env.reset()
-
-        while True:
-            # refresh env
-            env.render()
-
-            # RL choose action based on observation
-            action = RL.choose_action(observation)
-
-            # RL take action and get next observation and reward
-            observation_, reward, done = env.step(action)
-
-            # !! restore transition
-            RL.store_transition(observation, action, reward, observation_)
-
-            # 超过200条transition之后每隔5步学习一次
-            if (step > 200) and (step % 5 == 0):
-                RL.learn()
-
-            # swap observation
-            observation = observation_
-
-            # break while loop when end of this episode
-            if done:
-                break
-            step += 1
-
-    # end of game
-    print("game over")
-    env.destroy()
-
-
-if __name__ == "__main__":
-    # maze game
-    env = PartitionMazeEnv()
-
-    # TODO 代码还没有写完，跑不了！！！
-    RL = DeepQNetwork(env.n_actions, env.n_features,
-                      learning_rate=0.01,
-                      reward_decay=0.9,
-                      e_greedy=0.9,
-                      replace_target_iter=200,
-                      memory_size=2000)
-    env.after(100, run_maze)
-    env.mainloop()
-    RL.plot_cost()
-
-

Duel_Double_DQN/DQN.py

@@ -0,0 +1,144 @@
+import torch.nn.functional as F
+import torch.nn as nn
+import numpy as np
+import torch
+import copy
+
+
+def build_net(layer_shape, activation, output_activation):
+	'''Build networks with For loop'''
+	layers = []
+	for j in range(len(layer_shape)-1):
+		act = activation if j < len(layer_shape)-2 else output_activation
+		layers += [nn.Linear(layer_shape[j], layer_shape[j+1]), act()]
+	return nn.Sequential(*layers)
+
+class Q_Net(nn.Module):
+	def __init__(self, state_dim, action_dim, hid_shape):
+		super(Q_Net, self).__init__()
+		layers = [state_dim] + list(hid_shape) + [action_dim]
+		self.Q = build_net(layers, nn.ReLU, nn.Identity)
+	def forward(self, s):
+		q = self.Q(s)
+		return q
+
+
+class Duel_Q_Net(nn.Module):
+	def __init__(self, state_dim, action_dim, hid_shape):
+		super(Duel_Q_Net, self).__init__()
+		layers = [state_dim] + list(hid_shape)
+		self.hidden = build_net(layers, nn.ReLU, nn.ReLU)
+		self.V = nn.Linear(hid_shape[-1], 1)
+		self.A = nn.Linear(hid_shape[-1], action_dim)
+
+	def forward(self, s):
+		s = self.hidden(s)
+		Adv = self.A(s)
+		V = self.V(s)
+		Q = V + (Adv - torch.mean(Adv, dim=-1, keepdim=True))  # Q(s,a)=V(s)+A(s,a)-mean(A(s,a))
+		return Q
+
+
+class DQN_agent(object):
+	def __init__(self, **kwargs):
+		# Init hyperparameters for agent, just like "self.gamma = opt.gamma, self.lambd = opt.lambd, ..."
+		self.__dict__.update(kwargs)
+		self.tau = 0.005
+		self.replay_buffer = ReplayBuffer(self.state_dim, self.dvc, max_size=int(1e6))
+		if self.Duel:
+			self.q_net = Duel_Q_Net(self.state_dim, self.action_dim, (self.net_width,self.net_width)).to(self.dvc)
+		else:
+			self.q_net = Q_Net(self.state_dim, self.action_dim, (self.net_width, self.net_width)).to(self.dvc)
+		self.q_net_optimizer = torch.optim.Adam(self.q_net.parameters(), lr=self.lr)
+		self.q_target = copy.deepcopy(self.q_net)
+		# Freeze target networks with respect to optimizers (only update via polyak averaging)
+		for p in self.q_target.parameters(): p.requires_grad = False
+
+
+	def select_action(self, state, deterministic):#only used when interact with the env
+		with torch.no_grad():
+			state = torch.FloatTensor(state.reshape(1, -1)).to(self.dvc)
+			# if deterministic:
+			# 	a = self.q_net(state).argmax().item()
+			# else:
+			if np.random.rand() < self.exp_noise:
+				if state[0][0] == 0:
+					a = np.random.randint(0,10)
+				else:
+					a = np.random.randint(10,13)
+			else:
+				if state[0][0] == 0:
+					q_value = self.q_net(state)
+					q_value[:10] = - float('inf')
+					a = q_value.argmax().item()
+				else:
+					q_value = self.q_net(state)
+					q_value[10:] = - float('inf')
+					a = q_value.argmax().item()
+		return a
+
+
+	def train(self):
+		s, a, r, s_next, dw = self.replay_buffer.sample(self.batch_size)
+
+		'''Compute the target Q value'''
+		with torch.no_grad():
+			if self.Double:
+				argmax_a = self.q_net(s_next).argmax(dim=1).unsqueeze(-1)
+				max_q_next = self.q_target(s_next).gather(1,argmax_a)
+			else:
+				max_q_next = self.q_target(s_next).max(1)[0].unsqueeze(1)
+			target_Q = r + (~dw) * self.gamma * max_q_next #dw: die or win
+
+		# Get current Q estimates
+		current_q = self.q_net(s)
+		current_q_a = current_q.gather(1,a)
+
+		q_loss = F.mse_loss(current_q_a, target_Q)
+		self.q_net_optimizer.zero_grad()
+		q_loss.backward()
+		self.q_net_optimizer.step()
+
+		# Update the frozen target models
+		for param, target_param in zip(self.q_net.parameters(), self.q_target.parameters()):
+			target_param.data.copy_(self.tau * param.data + (1 - self.tau) * target_param.data)
+
+
+	def save(self,algo,EnvName,steps):
+		torch.save(self.q_net.state_dict(), "./weights/{}_{}_{}.pth".format(algo,EnvName,steps))
+
+	def load(self,algo,EnvName,steps):
+		self.q_net.load_state_dict(torch.load("./model/{}_{}_{}.pth".format(algo,EnvName,steps),map_location=self.dvc))
+		self.q_target.load_state_dict(torch.load("./model/{}_{}_{}.pth".format(algo,EnvName,steps),map_location=self.dvc))
+
+
+class ReplayBuffer(object):
+	def __init__(self, state_dim, dvc, max_size=int(1e6)):
+		self.max_size = max_size
+		self.dvc = dvc
+		self.ptr = 0
+		self.size = 0
+
+		self.s = torch.zeros((max_size, state_dim),dtype=torch.float,device=self.dvc)
+		self.a = torch.zeros((max_size, 1),dtype=torch.long,device=self.dvc)
+		self.r = torch.zeros((max_size, 1),dtype=torch.float,device=self.dvc)
+		self.s_next = torch.zeros((max_size, state_dim),dtype=torch.float,device=self.dvc)
+		self.dw = torch.zeros((max_size, 1),dtype=torch.bool,device=self.dvc)
+
+	def add(self, s, a, r, s_next, dw):
+		self.s[self.ptr] = torch.from_numpy(s).to(self.dvc)
+		self.a[self.ptr] = a
+		self.r[self.ptr] = r
+		self.s_next[self.ptr] = torch.from_numpy(s_next).to(self.dvc)
+		self.dw[self.ptr] = dw
+
+		self.ptr = (self.ptr + 1) % self.max_size
+		self.size = min(self.size + 1, self.max_size)
+
+	def sample(self, batch_size):
+		ind = torch.randint(0, self.size, device=self.dvc, size=(batch_size,))
+		return self.s[ind], self.a[ind], self.r[ind], self.s_next[ind], self.dw[ind]
+
+
+
+

Duel_Double_DQN/main.py

@@ -0,0 +1,163 @@
+import gymnasium as gym
+import os
+import shutil
+import argparse
+import torch
+import sys
+sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+from env_dis import PartitionMazeEnv
+from utils import evaluate_policy, str2bool
+from datetime import datetime
+from DQN import DQN_agent
+
+'''Hyperparameter Setting'''
+parser = argparse.ArgumentParser()
+parser.add_argument('--dvc', type=str, default='cpu',
+                    help='running device: cuda or cpu')
+parser.add_argument('--EnvIdex', type=int, default=0, help='CP-v1, LLd-v2')
+parser.add_argument('--write', type=str2bool, default=False,
+                    help='Use SummaryWriter to record the training')
+parser.add_argument('--render', type=str2bool,
+                    default=False, help='Render or Not')
+parser.add_argument('--Loadmodel', type=str2bool,
+                    default=False, help='Load pretrained model or Not')
+parser.add_argument('--ModelIdex', type=int, default=100,
+                    help='which model to load')
+
+parser.add_argument('--seed', type=int, default=42, help='random seed')
+parser.add_argument('--Max_train_steps', type=int,
+                    default=int(1e8), help='Max training steps')
+parser.add_argument('--save_interval', type=int,
+                    default=int(50e3), help='Model saving interval, in steps.')
+parser.add_argument('--eval_interval', type=int, default=int(2e3),
+                    help='Model evaluating interval, in steps.')
+parser.add_argument('--random_steps', type=int, default=int(3e3),
+                    help='steps for random policy to explore')
+parser.add_argument('--update_every', type=int,
+                    default=50, help='training frequency')
+
+parser.add_argument('--gamma', type=float, default=0.99,
+                    help='Discounted Factor')
+parser.add_argument('--net_width', type=int,
+                    default=200, help='Hidden net width')
+parser.add_argument('--lr', type=float, default=1e-4, help='Learning rate')
+parser.add_argument('--batch_size', type=int, default=256,
+                    help='lenth of sliced trajectory')
+parser.add_argument('--exp_noise', type=float,
+                    default=0.2, help='explore noise')
+parser.add_argument('--noise_decay', type=float, default=0.99,
+                    help='decay rate of explore noise')
+parser.add_argument('--Double', type=str2bool, default=True,
+                    help='Whether to use Double Q-learning')
+parser.add_argument('--Duel', type=str2bool, default=True,
+                    help='Whether to use Duel networks')
+opt = parser.parse_args()
+opt.dvc = torch.device(opt.dvc)  # from str to torch.device
+print(opt)
+
+
+def main():
+    EnvName = ['CartPole-v1', 'LunarLander-v2']
+    BriefEnvName = ['PM_DQN', 'CPV1', 'LLdV2']
+    # env = gym.make(EnvName[opt.EnvIdex], render_mode = "human" if opt.render else None)
+    # eval_env = gym.make(EnvName[opt.EnvIdex])
+    env = PartitionMazeEnv()
+    eval_env = PartitionMazeEnv()
+    opt.state_dim = env.observation_space.shape[0]
+    opt.action_dim = env.action_space.n
+    opt.max_e_steps = 50
+
+    # Algorithm Setting
+    if opt.Duel:
+        algo_name = 'Duel'
+    else:
+        algo_name = ''
+    if opt.Double:
+        algo_name += 'DDQN'
+    else:
+        algo_name += 'DQN'
+
+    # Seed Everything
+    env_seed = opt.seed
+    torch.manual_seed(opt.seed)
+    torch.cuda.manual_seed(opt.seed)
+    torch.backends.cudnn.deterministic = True
+    torch.backends.cudnn.benchmark = False
+    print("Random Seed: {}".format(opt.seed))
+
+    print('Algorithm:', algo_name, '  Env:', BriefEnvName[opt.EnvIdex], '  state_dim:', opt.state_dim,
+          '  action_dim:', opt.action_dim, '  Random Seed:', opt.seed, '  max_e_steps:', opt.max_e_steps, '\n')
+
+    if opt.write:
+        from torch.utils.tensorboard import SummaryWriter
+        timenow = str(datetime.now())[0:-10]
+        timenow = ' ' + timenow[0:13] + '_' + timenow[-2::]
+        writepath = 'runs/{}-{}_S{}_'.format(algo_name,
+                                             BriefEnvName[opt.EnvIdex], opt.seed) + timenow
+        if os.path.exists(writepath):
+            shutil.rmtree(writepath)
+        writer = SummaryWriter(log_dir=writepath)
+
+    # Build model and replay buffer
+    if not os.path.exists('model'):
+        os.mkdir('model')
+    agent = DQN_agent(**vars(opt))
+    if opt.Loadmodel:
+        agent.load(algo_name, BriefEnvName[opt.EnvIdex], opt.ModelIdex)
+
+    if opt.render:
+        while True:
+            score = evaluate_policy(env, agent, 1)
+            print('EnvName:', BriefEnvName[opt.EnvIdex],
+                  'seed:', opt.seed, 'score:', score)
+    else:
+        total_steps = 0
+        while total_steps < opt.Max_train_steps:
+            # Do not use opt.seed directly, or it can overfit to opt.seed
+            s = env.reset(seed=env_seed)
+            env_seed += 1
+            done = False
+
+            '''Interact & trian'''
+            while not done:
+                # e-greedy exploration
+                if total_steps < opt.random_steps:
+                    a = env.action_space.sample()
+                else:
+                    a = agent.select_action(s, deterministic=False)
+                s_next, r, dw, tr, info = env.step(a)
+                done = (dw or tr)
+
+                agent.replay_buffer.add(s, a, r, s_next, dw)
+                s = s_next
+
+                '''Update'''
+                # train 50 times every 50 steps rather than 1 training per step. Better!
+                if total_steps >= opt.random_steps and total_steps % opt.update_every == 0:
+                    for j in range(opt.update_every):
+                        agent.train()
+
+                '''Noise decay & Record & Log'''
+                if total_steps % 1000 == 0:
+                    agent.exp_noise *= opt.noise_decay
+                if total_steps % opt.eval_interval == 0:
+                    score = evaluate_policy(eval_env, agent, turns=3)
+                    if opt.write:
+                        writer.add_scalar(
+                            'ep_r', score, global_step=total_steps)
+                        writer.add_scalar(
+                            'noise', agent.exp_noise, global_step=total_steps)
+                    print('EnvName:', BriefEnvName[opt.EnvIdex], 'seed:', opt.seed, 'steps: {}k'.format(
+                        int(total_steps/1000)), 'score:', int(score))
+                total_steps += 1
+
+                '''save model'''
+                if total_steps % opt.save_interval == 0:
+                    agent.save(algo_name, BriefEnvName[opt.EnvIdex], int(
+                        total_steps/1000))
+    env.close()
+    eval_env.close()
+
+
+if __name__ == '__main__':
+    main()

Duel_Double_DQN/utils.py

@@ -0,0 +1,28 @@
+def evaluate_policy(env, agent, turns = 3):
+    total_scores = 0
+    for j in range(turns):
+        s = env.reset()
+        done = False
+        while not done:
+            # Take deterministic actions at test time
+            a = agent.select_action(s, deterministic=True)
+            s_next, r, dw, tr, info = env.step(a)
+            done = (dw or tr)
+
+            total_scores += r
+            s = s_next
+    return int(total_scores/turns)
+
+
+#You can just ignore this funciton. Is not related to the RL.
+def str2bool(v):
+    '''transfer str to bool for argparse'''
+    if isinstance(v, bool):
+        return v
+    if v.lower() in ('yes', 'True','true','TRUE', 't', 'y', '1'):
+        return True
+    elif v.lower() in ('no', 'False','false','FALSE', 'f', 'n', '0'):
+        return False
+    else:
+        print('Wrong Input.')
+        raise

env.py

@@ -39,8 +39,8 @@ class PartitionMazeEnv(gym.Env):
         ##############################
         # 可能需要手动修改的超参数
         ##############################
-        self.CUT_NUM = 2    # 横切一半，竖切一半
+        self.CUT_NUM = 6    # 横切一半，竖切一半
-        self.BASE_LINE = 4000     # 基准时间，通过greedy或者蒙特卡洛计算出来
+        self.BASE_LINE = 12000     # 基准时间，通过greedy或者蒙特卡洛计算出来
 
         self.phase = 0    # 阶段控制，0：区域划分阶段，1：迷宫初始化阶段，2：走迷宫阶段
         self.partition_step = 0      # 区域划分阶段步数，范围 0~4

env_dis.py

@@ -0,0 +1,278 @@
+import gymnasium as gym
+from gymnasium import spaces
+import numpy as np
+import yaml
+import math
+
+
+class PartitionMazeEnv(gym.Env):
+    """
+    自定义环境，分为两阶段：
+    阶段 0：区域切分（共 4 步，每一步输出一个标量，用于确定竖切和横切位置）。
+           切分顺序为：第一步输出 c₁，第二步输出 c₂，第三步输出 r₁，第四步输出 r₂。
+           离散化后取值仅为 {0, 0.1, 0.2, …, 0.9}（其中 0 表示不切）。
+    阶段 1：车辆路径规划（走迷宫），车辆从区域中心出发，在九宫格内按照上下左右移动，
+           直到所有目标格子被覆盖或步数上限达到。
+    """
+
+    def __init__(self, config=None):
+        super(PartitionMazeEnv, self).__init__()
+        # 车队参数设置
+        with open('params.yml', 'r', encoding='utf-8') as file:
+            params = yaml.safe_load(file)
+
+        self.H = params['H']
+        self.W = params['W']
+        self.num_cars = params['num_cars']
+
+        self.flight_time_factor = params['flight_time_factor']
+        self.comp_time_factor = params['comp_time_factor']
+        self.trans_time_factor = params['trans_time_factor']
+        self.car_time_factor = params['car_time_factor']
+        self.bs_time_factor = params['bs_time_factor']
+
+        self.flight_energy_factor = params['flight_energy_factor']
+        self.comp_energy_factor = params['comp_energy_factor']
+        self.trans_energy_factor = params['trans_energy_factor']
+        self.battery_energy_capacity = params['battery_energy_capacity']
+
+        ##############################
+        # 可能需要手动修改的超参数
+        ##############################
+        self.CUT_NUM = 4    # 横切一半，竖切一半
+        self.BASE_LINE = 4000     # 基准时间，通过greedy或者蒙特卡洛计算出来
+
+        self.phase = 0    # 阶段控制，0：区域划分阶段，1：迷宫初始化阶段，2：走迷宫阶段
+        self.partition_step = 0      # 区域划分阶段步数，范围 0~4
+        self.partition_values = np.zeros(
+            self.CUT_NUM, dtype=np.float32)  # 存储 c₁, c₂, r₁, r₂
+
+        # 定义动作空间：长度为 14 的离散动作空间
+        # 前 10 个表示切分动作 {0, 0.1, ..., 0.9}，后 4 个表示上下左右移动
+        self.action_space = spaces.Discrete(14)
+
+        # 定义观察空间为8维向量
+        # TODO 返回的状态目前只有位置坐标
+        # 阶段 0 状态：前 4 维表示已决策的切分值（未决策部分为 0）
+        # 阶段 1 状态：车辆位置 (2D)
+        self.observation_space = spaces.Box(
+            low=0.0, high=1.0, shape=(1 + self.CUT_NUM + 2 * self.num_cars,), dtype=np.float32)
+
+        # 切分阶段相关变量
+        self.col_cuts = []     # 存储竖切位置（c₁, c₂），当值为0时表示不切
+        self.row_cuts = []   # 存储横切位置（r₁, r₂）
+
+        self.init_maze_step = 0
+
+        # 路径规划阶段相关变量
+        self.MAX_STEPS = 50         # 迷宫走法步数上限
+        self.step_count = 0
+        self.rectangles = {}
+        self.car_pos = [(self.H / 2, self.W / 2) for _ in range(self.num_cars)]
+        self.car_traj = [[] for _ in range(self.num_cars)]
+        self.current_car_index = 0
+
+    def reset(self, seed=None, options=None):
+        # 重置所有变量，回到切分阶段（phase 0）
+        self.phase = 0
+        self.partition_step = 0
+        self.partition_values = np.zeros(self.CUT_NUM, dtype=np.float32)
+        self.col_cuts = []
+        self.row_cuts = []
+        self.init_maze_step = 0
+        self.region_centers = []
+        self.step_count = 0
+        self.rectangles = {}
+        self.car_pos = [(self.H / 2, self.W / 2) for _ in range(self.num_cars)]
+        self.car_traj = [[] for _ in range(self.num_cars)]
+        self.current_car_index = 0
+        # 状态：前 4 维为 partition_values，其余补 0
+        state = np.concatenate(
+            [[self.phase], self.partition_values, np.zeros(np.array(self.car_pos).flatten().shape[0], dtype=np.float32)])
+        return state
+
+    def step(self, action):
+        # 在所有阶段动作均为离散动作
+        if self.phase == 0:
+            # 切分阶段：前 10 个动作对应 {0, 0.1, ..., 0.9}
+            disc_val = action * 0.1  # 修正为动作直接映射到切分比例
+            self.partition_values[self.partition_step] = disc_val
+            self.partition_step += 1
+
+            # 构造当前状态：前 partition_step 个为已决策值，其余为 0，再补 7 个 0
+            state = np.concatenate(
+                [[self.phase], self.partition_values, np.zeros(
+                    np.array(self.car_pos).flatten().shape[0], dtype=np.float32)]
+            )
+
+            # 如果未完成 4 步，则仍处于切分阶段，不发奖励，done 为 False
+            if self.partition_step < self.CUT_NUM:
+                return state, 0.0, False, False, {}
+            else:
+                # 完成 4 步后，计算切分边界
+                # 过滤掉 0，并去重后排序
+                vert = sorted(set(v for v in self.partition_values[:len(
+                    self.partition_values) // 2] if v > 0))
+                horiz = sorted(set(v for v in self.partition_values[len(
+                    self.partition_values) // 2:] if v > 0))
+                vertical_cuts = vert if vert else []
+                horizontal_cuts = horiz if horiz else []
+
+                # 边界：始终包含 0 和 1
+                self.col_cuts = [0.0] + vertical_cuts + [1.0]
+                self.row_cuts = [0.0] + horizontal_cuts + [1.0]
+
+                # 判断分区是否合理，并计算各个分区的任务卸载率ρ
+                valid_partition = True
+                for i in range(len(self.row_cuts) - 1):
+                    for j in range(len(self.col_cuts) - 1):
+                        d = (self.col_cuts[j+1] - self.col_cuts[j]) * self.W * \
+                            (self.row_cuts[i+1] - self.row_cuts[i]) * self.H
+                        rho_time_limit = (self.flight_time_factor - self.trans_time_factor) / \
+                            (self.comp_time_factor - self.trans_time_factor)
+                        rho_energy_limit = (self.battery_energy_capacity - self.flight_energy_factor * d - self.trans_energy_factor * d) / \
+                            (self.comp_energy_factor * d -
+                             self.trans_energy_factor * d)
+                        if rho_energy_limit < 0:
+                            valid_partition = False
+                            break
+                        rho = min(rho_time_limit, rho_energy_limit)
+
+                        flight_time = self.flight_time_factor * d
+                        bs_time = self.bs_time_factor * (1 - rho) * d
+
+                        self.rectangles[(i, j)] = {
+                            'center': ((self.row_cuts[i] + self.row_cuts[i+1]) * self.H / 2, (self.col_cuts[j+1] + self.col_cuts[j]) * self.W / 2),
+                            'flight_time': flight_time,
+                            'bs_time': bs_time,
+                            'is_visited': False
+                        }
+                    if not valid_partition:
+                        break
+
+                if not valid_partition:
+                    reward = -10000
+                    state = np.concatenate(
+                        [[self.phase], self.partition_values, np.zeros(np.array(self.car_pos).flatten().shape[0], dtype=np.float32)])
+                    return state, reward, True, False, {}
+                else:
+                    # 进入阶段 1：初始化迷宫
+                    self.phase = 1
+                    state = np.concatenate(
+                        [[self.phase], self.partition_values, np.array(self.car_pos).flatten()])
+                    reward = 10
+
+                    # 构建反向索引，方便后续计算
+                    self.reverse_rectangles = {
+                        v['center']: k for k, v in self.rectangles.items()}
+                    return state, reward, False, False, {}
+
+        elif self.phase == 1:
+            # TODO 阶段一可以不写出来！！！
+            # 阶段 1：初始化迷宫，让多个车辆从区域中心出发，前往最近的几个区域中心点
+            region_centers = [
+                (i, j, self.rectangles[(i, j)]['center'])
+                for i in range(len(self.row_cuts) - 1)
+                for j in range(len(self.col_cuts) - 1)
+            ]
+            # 按照与区域中心的距离从近到远排序
+            region_centers.sort(
+                key=lambda x: math.dist(x[2], (self.H / 2, self.W / 2))
+            )
+
+            # 分配最近的区域给每辆车
+            for idx in range(self.num_cars):
+                i, j, center = region_centers[idx]
+                self.car_pos[idx] = center
+                self.car_traj[idx].append((i, j))
+                self.rectangles[(i, j)]['is_visited'] = True
+
+            # 进入阶段 2：走迷宫
+            self.phase = 2
+            state = np.concatenate(
+                [[self.phase], self.partition_values,
+                    np.array(self.car_pos).flatten()]
+            )
+            return state, 0.0, False, False, {}
+
+        elif self.phase == 2:
+            # 阶段 2：路径规划（走迷宫）
+            # 后 4 个动作对应上下左右移动
+            current_car = self.current_car_index
+            current_row, current_col = self.reverse_rectangles[self.car_pos[current_car]]
+
+            # 初始化新的行、列为当前值
+            new_row, new_col = current_row, current_col
+
+            if action == 10 and current_row > 0:  # 上
+                new_row = current_row - 1
+            elif action == 11 and current_row < len(self.row_cuts) - 2:  # 下
+                new_row = current_row + 1
+            elif action == 12 and current_col > 0:  # 左
+                new_col = current_col - 1
+            elif action == 13 and current_col < len(self.col_cuts) - 2:  # 右
+                new_col = new_col + 1
+
+            # 更新车辆位置
+            self.car_pos[current_car] = self.rectangles[(
+                new_row, new_col)]['center']
+            if new_row != current_row or new_col != current_col:
+                self.car_traj[current_car].append((new_row, new_col))
+            self.step_count += 1
+            self.current_car_index = (
+                self.current_car_index + 1) % self.num_cars
+
+            # 更新访问标记：将新网格标记为已访问
+            self.rectangles[(new_row, new_col)]['is_visited'] = True
+
+            # 观察状态
+            state = np.concatenate(
+                [[self.phase], self.partition_values, np.array(self.car_pos).flatten()])
+            reward = 0
+
+            # Episode 终止条件：所有网格均被访问或步数达到上限
+            done = all([value['is_visited'] for _, value in self.rectangles.items()]) or (
+                self.step_count >= self.MAX_STEPS)
+            if done and all([value['is_visited'] for _, value in self.rectangles.items()]):
+                # 区域覆盖完毕，根据轨迹计算各车队的执行时间
+                T = max([self._compute_motorcade_time(idx)
+                        for idx in range(self.num_cars)])
+                # print(T)
+                # print(self.partition_values)
+                # print(self.car_traj)
+                reward += self.BASE_LINE / T * 100
+            elif done and self.step_count >= self.MAX_STEPS:
+                reward += -1000
+
+            return state, reward, done, False, {}
+
+    def _compute_motorcade_time(self, idx):
+        flight_time = sum(self.rectangles[tuple(point)]['flight_time']
+                          for point in self.car_traj[idx])
+        bs_time = sum(self.rectangles[tuple(point)]['bs_time']
+                      for point in self.car_traj[idx])
+
+        # 计算车的移动时间，首先在轨迹的首尾添加上大区域中心
+        car_time = 0
+        for i in range(len(self.car_traj[idx]) - 1):
+            first_point = self.car_traj[idx][i]
+            second_point = self.car_traj[idx][i + 1]
+            car_time += math.dist(self.rectangles[first_point]['center'], self.rectangles[second_point]['center']) * \
+                self.car_time_factor
+        car_time += math.dist(self.rectangles[self.car_traj[idx][0]]['center'], [
+                              self.H / 2, self.W / 2]) * self.car_time_factor
+        car_time += math.dist(self.rectangles[self.car_traj[idx][-1]]['center'], [
+                              self.H / 2, self.W / 2]) * self.car_time_factor
+
+        return max(float(car_time) + flight_time, bs_time)
+
+    def render(self):
+        if self.phase == 1:
+            print("Phase 1: Initialize maze environment.")
+            print(f"Partition values so far: {self.partition_values}")
+            print(f"Motorcade positon: {self.car_pos}")
+            # input('1111')
+        elif self.phase == 2:
+            print("Phase 2: Play maze.")
+            print(f'Motorcade trajectory: {self.car_traj}')
+            # input('2222')