0

我正在使用带有 BipedalWalker-v3 环境的 python gym。设置 DQNAgent 时出现以下错误

IndexError                                Traceback (most recent call last)
~\AppData\Local\Temp/ipykernel_7456/1216894816.py in <module>
----> 1 dqn.fit(env, nb_steps=1000, visualize=False, verbose=1)

~\.conda\envs\AI\lib\site-packages\rl\core.py in fit(self, env, nb_steps, action_repetition, callbacks, verbose, visualize, nb_max_start_steps, start_step_policy, log_interval, nb_max_episode_steps)
    174                 for _ in range(action_repetition):
    175                     callbacks.on_action_begin(action)
--> 176                     observation, r, done, info = env.step(action)
    177                     observation = deepcopy(observation)
    178                     if self.processor is not None:

~\.conda\envs\AI\lib\site-packages\gym\wrappers\time_limit.py in step(self, action)
     16             self._elapsed_steps is not None
     17         ), "Cannot call env.step() before calling reset()"
---> 18         observation, reward, done, info = self.env.step(action)
     19         self._elapsed_steps += 1
     20         if self._elapsed_steps >= self._max_episode_steps:

~\.conda\envs\AI\lib\site-packages\gym\envs\box2d\bipedal_walker.py in step(self, action)
    413             print('Action inpuit type: {}'.format(type(action)))
    414             #print(action[0])
--> 415             self.joints[0].motorSpeed = float(SPEED_HIP * np.sign(action[0]))
    416             self.joints[0].maxMotorTorque = float(
    417                 MOTORS_TORQUE * np.clip(np.abs(action[0]), 0, 1)

IndexError: invalid index to scalar variable.

我在gym 包内的.step() 函数中添加了一些打印函数（就在else: 行的下方），以查看我为action 变量获得了什么输入。

def step(self, action):
        # self.hull.ApplyForceToCenter((0, 20), True) -- Uncomment this to receive a bit of stability help
        control_speed = False  # Should be easier as well
        if control_speed:
            self.joints[0].motorSpeed = float(SPEED_HIP * np.clip(action[0], -1, 1))
            self.joints[1].motorSpeed = float(SPEED_KNEE * np.clip(action[1], -1, 1))
            self.joints[2].motorSpeed = float(SPEED_HIP * np.clip(action[2], -1, 1))
            self.joints[3].motorSpeed = float(SPEED_KNEE * np.clip(action[3], -1, 1))
        else:
            print('Action inpuit: {}'.format(action))
            print('Action inpuit type: {}'.format(type(action)))
            #print(action[0])
            self.joints[0].motorSpeed = float(SPEED_HIP * np.sign(action[0]))
            self.joints[0].maxMotorTorque = float(
                MOTORS_TORQUE * np.clip(np.abs(action[0]), 0, 1)
            )
            self.joints[1].motorSpeed = float(SPEED_KNEE * np.sign(action[1]))
            self.joints[1].maxMotorTorque = float(
                MOTORS_TORQUE * np.clip(np.abs(action[1]), 0, 1)
            )
            self.joints[2].motorSpeed = float(SPEED_HIP * np.sign(action[2]))
            self.joints[2].maxMotorTorque = float(
                MOTORS_TORQUE * np.clip(np.abs(action[2]), 0, 1)
            )
            self.joints[3].motorSpeed = float(SPEED_KNEE * np.sign(action[3]))
            self.joints[3].maxMotorTorque = float(
                MOTORS_TORQUE * np.clip(np.abs(action[3]), 0, 1)
            )

我得到以下输出：

训练1000步...

动作输入：[0 0 0 0]

动作输入类型：<class 'numpy.ndarray'>

间隔 1（执行 0 步）

动作输入：3

动作输入类型：<class 'numpy.int32'>

（索引错误）

从我可以看到第二次将动作变量提供给健身房环境步骤函数，它是以 int 的形式而不是应有的数组形式。

对此的任何帮助将不胜感激。

完整代码如下：

import gym 
import random
import numpy as np
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Flatten
from tensorflow.keras.optimizers import Adam
from rl.agents import DQNAgent
from rl.policy import BoltzmannQPolicy
from rl.memory import SequentialMemory

def build_model(states, actions):
    model = Sequential()
    model.add(Flatten(input_shape=(1,states)))
    model.add(Dense(24, activation='relu'))
    model.add(Dense(24, activation='relu'))
    model.add(Dense(actions, activation='linear'))
    return model

def build_agent(model, actions):
    policy = BoltzmannQPolicy()
    memory = SequentialMemory(limit=50000, window_length=1)
    dqn = DQNAgent(model=model, memory=memory, policy=policy, 
                   nb_actions=actions, nb_steps_warmup=10, target_model_update=1e-2)
    return dqn

env = gym.make('BipedalWalker-v3')
#env = gym.make('CartPole-v1')
states = env.observation_space.shape[0]
actions = env.action_space.shape[0]
#actions = env.action_space.n

model = build_model(states, actions)
dqn = build_agent(model, actions)
dqn.compile(Adam(lr=1e-3), metrics=['mae'])
dqn.fit(env, nb_steps=1000, visualize=False, verbose=1)
4

0 回答 0