OpenAI gym-like reinforcement environment created by V-REP
- V-REP (3.5 or later)
- Its directory name should be changed to V-REP_PRO_EDU.
- Its directory should be on ...
- ... home directory (~/) in the case of linux.
- ... Applications (/Applications/) in the case of mac.
- numpy
- gym
- fasteners
- git-lfs
git lfs clone https://github.com/kbys-t/gym_vrep.git
cd gym_vrep
pip install -e .For example,
import gym_vrep
env = gym_vrep.VrepEnv(scene=env_name, is_render=is_check, is_boot=is_boot, port=19997)
if is_record:
env.monitor(save_dir, force=True)
for epi in range(n_episode):
observation = env.reset()
for stp in range(n_time):
observation, reward, done, info = env.step(action)
env.close()- By changing port number, multiple simulations can be processed.
- Please be aware that the specified port is opened at the OS level.
- If vrep process is already running, please set
is_boot=False.- In that case, a scene file is not required to be opened.