WuXinyang2012

Records of drl-for-safety projects.

1, Background Kownledge.

1.1 Learn the basic DRL concepts and popular algorithms.
1.2 Focus more on DDPG+HER algorithm.

2, Compare mujoco-py and dm_control.

Result: Choose mujoco-py as our framework.

Reason:

1.) mujoco-py has more active community: the number of open issues ~120, while dm_control ~10.

WuXinyang2012

#!/usr/bin/env python

import os
import mujoco_py
import numpy as np


PATH_TO_HUMANOID_XML = os.path.expanduser('~/.mujoco/mjpro150/model/humanoid.xml')
#PATH_TO_HUMANOID_XML = os.path.expanduser('/home/rfa-xw/anaconda3/envs/mujoco-py/lib/python3.6/site-packages/gym/envs/robotics/assets/fetch/reach.xml')

WuXinyang2012

###Mac OS ignore
.DS_Store

### PyCharm ###

# User-specific stuff
.idea/**/workspace.xml
.idea/**/tasks.xml
.idea/**/usage.statistics.xml
.idea/**/dictionaries

WuXinyang2012

1.)qpos: position; qvel: velocity

2.)mujoco-py use actuator to make the move
joint type: free, ball, slide, hinge actuator type: motor, position, velocity, cylinder, muscle sensor type: touch, accelerometer, velocimeter, gyro, force, torque, magnetometer...

3.)

WuXinyang2012

"""
Cart pole swing-up: Identical version to PILCO V0.9
"""

import logging
import math
import gym
from gym import spaces
from gym.utils import seeding
import numpy as np