DQN

[guest-oo]

class mazeenv(gym.Env): def init(self): self.action_space = spaces.Discrete(8) self.observation_space = spaces.Box(low=0, high=23, shape=(14,), dtype=np.float32) self.state = None self.seed() self.reset() def seed(self, seed=None): self.np_random, seed = seeding.np_random(seed) return [seed] def step(self, action): assert self.action_space.contains(action), "%r (%s) invalid" % ( action, type(action), ) aa=[[1,-1],[1,-1],[1,1],[-1,1],[1,0],[-1,0],[0,1],[0,-1]] length = len(self.state) n = 2 # 列数 m = length // n # 行数 self.state = np.reshape(self.state, (m, n)) self.ps=self.state[0] self.goal=self.state[1] position = copy.deepcopy(self.ps) position[0] = position[0]+aa[action][0] position[1] = position[1]+aa[action][1] self.obs = [[2, 2], [3, 3], [5, 5], [7, 9], [11, 2]] if position[0]>=23 : reward = -100 position = copy.deepcopy(self.ps) done = False elif position[1]>=23 : reward = -100 position = copy.deepcopy(self.ps) done = False elif position[0]<0 : reward = -100 position = copy.deepcopy(self.ps) done = False elif position[1]<0 : reward = -100 position = copy.deepcopy(self.ps) done = False elif any(np.all(position == o) for o in self.obs): reward = -100 position = copy.deepcopy(self.ps) done = False elif position[0] == self.goal[0] and position[1] == self.goal[1]: reward = 100 done = True else: reward=-1 done = False position=np.array(position) self.state[0] = position self.state = np.reshape(self.state, (-1,)) return np.array(self.state), reward, done, {} def reset(self): #self.start = self.np_random.randint(0, 23, size=(1, 2)) self.obs = np.array([[2, 2], [3, 3], [5, 5], [7, 9], [11, 2]]) #self.goal = self.np_random.randint(0, 23, size=(1, 2)) self.start=np.array([[1,1]]) self.goal=np.array([[22,22]]) # 检查 self.state 是否与 obs 中的任何一个元素相等 while any(np.array_equal(self.start, o) for o in self.obs) or self.start[0, 0] < 0 or self.start[0, 1] < 0 or
self.start[0, 0] >= 23 or self.start[0, 1] >= 23: self.start = np.random.randint(0, 23, size=(1, 2)) while any(np.array_equal(self.goal, o) for o in self.obs) or self.goal[0, 0] < 0 or self.goal[0, 1] < 0 or
self.goal[0, 0] >= 23 or self.goal[0, 1] >= 23 or np.array_equal(self.goal, self.start): self.goal = np.random.randint(0, 23, size=(1, 2)) self.state = np.concatenate((self.start, self.goal), axis=0) self.state=np.concatenate((self.state, self.obs), axis=0) self.state = np.reshape(self.state, (-1,)) return np.array(self.state)

我设计迷宫环境，在训练结束测试中一直找不到终点、。