強化學習基礎第一個環境（Maze）

1、本篇文章主要介紹在gym環境下env環境編寫（未待完續，空閒更新）

經典的迷宮寶藏例子

import numpy as np
import time
import tkinter as tk # 窗口界面庫

UNIT = 40   # 像素
MAZE_H = 5  # 高寬格子數
MAZE_W = 3

class Maze(tk.Tk): # 新類繼承父類tkinter.Tk
    def __init__(self):
        super(Maze, self).__init__() # super類繼承方法,初始化
        self.action_space = ['u', 'd', 'l', 'r']
        self.n_actions = len(self.action_space)
        self.title('maze')
        self.geometry('{0}x{1}'.format(MAZE_H * UNIT, MAZE_H * UNIT)) # format格式化函數 geometry分辨率函數
        self._build_maze()

    def _build_maze(self):
        self.canvas = tk.Canvas(self, bg='white', # tk畫圖組件
                           height=MAZE_H * UNIT,
                           width=MAZE_W * UNIT)
        for c in range(0, MAZE_W * UNIT, UNIT): # 網格
            x0, y0, x1, y1 = c, 0, c, MAZE_H * UNIT
            self.canvas.create_line(x0, y0, x1, y1)
        for r in range(0, MAZE_H * UNIT, UNIT):
            x0, y0, x1, y1 = 0, r, MAZE_W * UNIT, r
            self.canvas.create_line(x0, y0, x1, y1)

        origin = np.array([20, 20]) # 原點

        hell2_center = origin + np.array([UNIT * 0, UNIT * 1])
        self.hell2 = self.canvas.create_rectangle(
            hell2_center[0] - 15, hell2_center[1] - 15,
            hell2_center[0] + 15, hell2_center[1] + 15,
            fill='black')
        hell3_center = origin + np.array([UNIT * 1, UNIT * 1])
        self.hell3 = self.canvas.create_rectangle(
            hell3_center[0] - 15, hell3_center[1] - 15,
            hell3_center[0] + 15, hell3_center[1] + 15,
            fill='black')
        hell6_center = origin + np.array([UNIT * 1, UNIT * 3])
        self.hell6 = self.canvas.create_rectangle(
            hell6_center[0] - 15, hell6_center[1] - 15,
            hell6_center[0] + 15, hell6_center[1] + 15,
            fill='black')
        hell7_center = origin + np.array([UNIT * 2, UNIT * 3])
        self.hell7 = self.canvas.create_rectangle(
            hell7_center[0] - 15, hell7_center[1] - 15,
            hell7_center[0] + 15, hell7_center[1] + 15,
            fill='black')
        oval_center = origin + np.array([UNIT * 2, UNIT * 4]) # 寶藏
        self.oval = self.canvas.create_oval(
            oval_center[0] - 15, oval_center[1] - 15,
            oval_center[0] + 15, oval_center[1] + 15,
            fill='yellow')
        self.rect = self.canvas.create_rectangle( # 主體
            origin[0] - 15, origin[1] - 15,
            origin[0] + 15, origin[1] + 15,
            fill='red')

        self.canvas.pack() # 打包

    def reset(self): 
        self.update()
        time.sleep(0.1)
        self.canvas.delete(self.rect) # 配置Python Tkinter 畫布（Canvas），刪除變化的矩形，然後重新創建
        origin = np.array([20, 20])
        self.rect = self.canvas.create_rectangle(
            origin[0] - 15, origin[1] - 15,
            origin[0] + 15, origin[1] + 15,
            fill='red') # 創建中心位置紅色方塊代表當前位置
        # return observation
        return self.canvas.coords(self.rect)

    def step(self, action):
        s = self.canvas.coords(self.rect)
        base_action = np.array([0, 0])
        if action == 0:   # up
            if s[1] > UNIT:
                base_action[1] -= UNIT
        elif action == 1:   # down
            if s[1] < (MAZE_H - 1) * UNIT:
                base_action[1] += UNIT
        elif action == 2:   # right
            if s[0] < (MAZE_W - 1) * UNIT:
                base_action[0] += UNIT
        elif action == 3:   # left
            if s[0] > UNIT:
                base_action[0] -= UNIT

        self.canvas.move(self.rect, base_action[0], base_action[1])  # move agent

        s_ = self.canvas.coords(self.rect)  # next state

        # reward function
        if s_ == self.canvas.coords(self.oval):
            reward = 1
            done = True
            s_ = 'terminal'
        elif s_ in [self.canvas.coords(self.hell2), self.canvas.coords(self.hell3),
                    self.canvas.coords(self.hell6), self.canvas.coords(self.hell7)]:
            reward = -1
            done = True
            s_ = 'terminal'
        else:
            reward = 0
            done = False

        return s_, reward, done

    def render(self):
        time.sleep(0.1)
        self.update()


def update():
    for t in range(10):
        s = env.reset()
        while True:
            env.render()
            a = 1
            s, r, done = env.step(a)
            if done:
                break

if __name__ == '__main__':
    env = Maze()
    env.after(100, update)
    env.mainloop()

經典的CartPole例子

reset函數：

    def reset(self):
        self.state = self.np_random.uniform(low=-0.05, high=0.05, size=(4,))
        # 利⽤均勻隨機分佈初試化環境的狀態
        self.steps_beyond_done = None
        # 設置當前步數爲None
        return np.array(self.state)

numpy的random庫， 用於生成隨機數矩陣，uniform函數用於產生均勻分佈隨機數（其他函數可以產生高斯分佈隨機數等。）

2、補充一點DQN算法模塊的函數閱讀

 函數：尋寶算法模塊choose_action

    def choose_action(self, observation):
        self.check_state_exist(observation)
        if np.random.uniform() < self.epsilon: # epsilon等於0.9，90%概率Q函數決定行動
            state_action = self.q_table.loc[observation, :] # 從Q表中提取觀測值那一行的Q值
            # 最大Q值對應多個動作時，隨機選擇動作
            action = np.random.choice(state_action[state_action == np.max(state_action)].index)
        else: # 10%概率隨機行動，探索其他可能
            # choose random action
            action = np.random.choice(self.actions)
        return action

注：這裏的第一個observation是[5,5,35,35]。這個數組矩陣來自於canvas.create_rectangle函數，根據左上和右下座標繪製矩形，定義一個像素40*40，則起點位置的狀態是[5,5,35,35]。

參考文獻：Pandas中loc和iloc函數用法詳解（源碼+實例） https://blog.csdn.net/w_weiying/article/details/81411257

參考文獻：莫煩老師，DQN代碼學習筆記  https://blog.csdn.net/yyyxxxsss/article/details/80467058