In the context of reinforcement learning, Mujoco + gym is more popular than Gazebo in research that involve robotics. However, mujoco-py released by OpenAI doesn’t provide full flexibility compared to original Mujoco C++ API. In a recent work of mine, I upgraded mujoco-py==1.5.0 that supports:
Interactive manipulation as provided by simulate in Mujoco, written in Cython
Force visualization similar to deepmind-control but allows for headless rendering
The code is available at https://github.com/davidsonic/self_brewed_mujoco_py
Based on self-brewed-mujoco-py, inverse kinmatics for Baxter robot was implemented, which given a snapshot of current scene, the robotic arm could automatically locate and perform grasping.
This is different with PickAndPlace task in latest gym environment, which requires reinforced PPO with millions of frames. My goal is to learn an adversary that would in turn assist robust robotic arm grasping.
A result of this procedure is available here.