How to set up a reinforcement learning environment for a UR5 robot?

asked 2019-04-30 07:47:37 -0500

bigum gravatar image

I would like to do some reinforcement learning on a Universal Robots UR5 in Gazebo simulation. Basically, I would like to make a path planner from an initial position to a position I choose. However, I'm a bit confused with the approach I should take. I can see there's both an openAI environment and a gym-gazebo environment. Which of these do I need to use? I don't see any implementations set up with a UR arm.

Currently I am able to control a UR5 in Gazebo through an action server (ur-gazebo package). But maybe that's not what I need really?

edit retag flag offensive close merge delete


Hi ! Are you able to do it ? I am also facing the same issue . Can you please share any resource or tell me how to set up the environment for that ?

Thanks in advance .

baleshwar gravatar image baleshwar  ( 2021-04-03 04:14:19 -0500 )edit