工作笔记｜基于tensorflow的强化学习小游戏CartPole-v0代码实现

游戏图像

训练结果

废话不说直接上代码

import numpy as np
import pygame
import math
import time
from load import *
from pygame.locals import *
import gym
import matplotlib.pyplot as plt
import tensorflow as tf
# import tensorflow.compat.v1 as tf
# tf.disable_v2_behavior()
# tf.compat.v1.disable_eager_execution()

# 摆脱Gym提供的仿真环境，自己创建CartPole仿真环境
class CartPoleEnv:
    def __init__(self):
        self.actions = [0,1]
        self.state = np.random.uniform(-0.05, 0.05, size=(4,))
        self.steps_beyond_done = 0
        self.viewer = None
        # 设置帧率
        self.FPSCLOCK = pygame.time.Clock()
        self.screen_size = [400,300]
        self.cart_x = 200
        self.cart_y = 200
        self.theta = -1.5
        self.gravity = 9.8
        self.mass_cart = 1.0
        self.mass_pole = 0.1
        self.total_mass = (self.mass_cart + self.mass_pole)
        self.length = 0.5
        self.pole_mess_length = (self.mass_pole * self.length)
        self.force_mag = 10.0
        self.tau = 0.02
    
    # 随机初始化
    def reset(self):
        n = np.random.randint(1, 1000, 1) 
        np.random.seed(n)
        self.state = np.random.uniform(-0.05, 0.05, size=(4,))
        self.steps_beyond_done = 0
        return np.array(self.state)
    
    def step(self, action):
        state = self.statex, x_dot, theta, theta_dot = stateforce = self.force_mag if action==1 else -self.force_mag
        costheta = math.cos(theta)
        sintheta = math.sin(theta)
        # 动力学方程
        temp = (force+self.pole_mess_length * theta_dot * theta_dot * sintheta)/self.total_mass
        thetaacc = (self.gravity * sintheta - costheta * temp)/(self.length * (4.0/3.0-self.mass_pole*costheta*costheta/self.total_mass))
        xacc = temp - self.pole_mess_length * thetaacc * costheta / self.total_mass
        x = x+self.tau * x_dot
        x_dot = x_dot + self.tau * xacc
        theta = theta +self.tau * theta_dot
        theta_dot = theta_dot * self.tau * thetaacc
        self.state