A toolkit for distributed reinforcement learning in robotics
My colleagues Matteo Lucchi and Friedemann Zindler have developed a toolkit for distributed reinforcement learning in robotics. With it we are able to let robots learn in simulation and transfer this to real robots with ease. The associated paper was accepted for presentation at IROS2020.
We have now release robo-gym on our JOANNEUM RESEARCH Github page.
Find an introductory blog post at ros discourse.