Baxter Control Environments

gym-baxter provides an interface for performing reinforcement learning and control experiments on robots in simulation and on real hardware. gym-baxter uses PyBullet as the backend for simulation experiments and ROS as the backend for hardware experiments. gym-baxter also makes use of the OpenAI Gym paradigm that is common for reinforcement learning environments.

Dependencies

Python 2.7
PyBullet
OpenAI Gym
ROS Kinetic
Baxter SDK
PyKDL
baxter_pykdl

Install

Install the Python dependencies:

pip install pybullet gym

To install gym-robotics:

git clone https://github.com/jmichaux/gym-baxter
cd gym-baxter
pip install -e .

To install ROS, follow the directions here. To create a workspace for Baxter, follow the directions here.

TODO

Baxter environments

• BaxterReacherEnv (in progress)
• BaxterPusherEnv (in progress)
• BaxterSliderEnv (in progress)
• BaxterPickPlaceEnv (in progress)