Reinforcement Learning

I am super new to simulators. I will need to implement a reinforcement learning algorithm on a robot so I wanted to learn Gazebo. Is there a comprehensive tutorial for using Gazebo with reinforcement. Any resource to get me on my way will be truly appreciated.

Check out OpenAI gym: 2019-tfm-ignacio-arranz/ at master · RoboticsLabURJC/2019-tfm-ignacio-arranz · GitHub . Just be aware that it uses “Gazebo Classic” (the older Gazebo) simulator, which is however more suitable for RL because it fully supports resetting the environment with a single service call (AFAIK the new Gazebo always struggled with this).

There’s also which uses the new Gazebo, but I don’t think it’s actively maintained anymore. As to the reset capability, I only see one issue with a particular system crashing after reset. Things should be working if you don’t use that system. @peci1, if you know of specific problems with reset, please let us know.