Roomba_MDP.py

import gym
import numpy 
impoort time
#Function for a random policy
def randomPolicy():
    return numpy.random.choice(4, size=((16)))
#Execution
def execute(env, policy, episode_len=100, render=False):
    reward = 0
    obs = env.reset()
    for t in range(episode_len):
        if render:
            env.render()
        action = policy[obs]
        obs, reward, done, _ = env.step(action)
        total_reward += reward
        if done:
            break
    return total_reward
#Evaluation
def optimalPolicy(env, policy, n_episodes=100):
    total_rewards = 0.0
    for _ in range(n_episodes):
        total_rewards += run_episode(env, policy)
    return total_rewards / n_episodes

if __name__ == '__main__':
    env = gym.make('FrozenLake-v0')
    ## Policy search
    maxIteration = 1000
    start = time.time()
    policy_set = [randomPolicy() for _ in range(maxIteration)]
    policy_score = [optimalPolicy(env, p) for p in policy_set]
    end = time.time()

    print("Best score = %0.2f. Time taken = %4.4f seconds" %(numpy.max(policy_score) , end - start))