ms, ma = env_model.sample_s_a() # ms in here is a str
mr, ms_ = env_model.get_r_s_(ms, ma)
RL.learn(ms, ma, mr, str(ms_))
# print(env_model.database)
# print('################')
# print(RL.q_table)
# print('################')
s = s_
s2 = s2_
if done:
s = env.reset()
break
if done2:
s2 = env.reset2()
break
# end of game
print('game over')
print(RL.q_table)
env.destroy()
if __name__ == "__main__":
env = Maze()
env_model = EnvModel(actions=list(range(env.n_actions)))
RL = Learning(actions=list(range(env.n_actions)))
env.after(0, update)
env.mainloop()
from env import Maze
from q_learning import QLearning
def update():
for episode in range(20):
state = env.reset()
step_count, done = 0, False
while not done:
env.render()
action = RL.choose_action(str(state))
state_, reward, done = env.step(action)
step_count += 1
RL.learn(str(state), action, reward, str(state_))
state = state_
print(' Round over at: {0} round, Total steps: {1} steps'.format(
episode, step_count))
if __name__ == '__main__':
env = Maze()
agent = QLearning(actions=list(range(env.n_actions)))
env.after(100, update())
# env.mainloop()
print('\n Q Table')
print(agent.q_table)
agent.q_table.to_csv('Q_Table.csv')
state_, reward, done = env.step(action)
step_count += 1 # 增加步数
# 机器人大脑从这个过渡（transition） (state, action, reward, state_) 中学习
RL.learn(str(state), action, reward, str(state_))
# 机器人移动到下一个 state
state = state_
# 如果踩到炸弹或者找到宝藏, 这回合就结束了
if done:
print("回合 {} 结束. 总步数 : {}\n".format(episode + 1, step_count))
break
# 结束游戏并关闭窗口
print('游戏结束')
env.destroy()
if __name__ == "__main__":
# 创建环境 env 和 RL
env = Maze()
RL = QLearning(actions=list(range(env.n_actions)))
# 开始可视化环境
env.after(100, update)
env.mainloop()
print('\nQ 表:')
print(RL.q_table)
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
time.sleep(60)
if __name__ == "__main__":
# maze game
env = Maze()
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,
# output_graph=True
)
env.after(100, run_maze)
env.mainloop()
RL.plot_cost()
