PyBullet-OpenAIGym-Baseline-ROS-CustomEnv

Code for Learning Agent and Environment Deployment

import gym
from baselines import deepq
import balance_bot

def callback(lcl, glb):
    # stop training if reward exceeds 199
    is_solved = lcl['t'] > 100 and sum(lcl['episode_rewards'][-101:-1]) / 100 >= 199
    return is_solved

def main():
    # create the environment
    env = gym.make("balancebot-v0") # <-- this we need to create

    # create the learning agent
    model = deepq.models.mlp([16, 16])

    # train the agent on the environment
    act = deepq.learn(
        env,
        #q_func=model,
        lr=1e-3,
        total_timesteps=100000,
        buffer_size=100000,
        exploration_fraction=0.1,
        exploration_final_eps=0.02,
        print_freq=10,
        callback=callback,
         network='mlp',
    )

    # save trained model
    act.save("balance.pkl")

if __name__ == '__main__':
    main()