Sai Krishna Gottipati
TalkRL: The Reinforcement Learning Podcast
00:00
Asymmetric Self Play for Automatic Gold Discovery in Robotic Manipulation
The paper was a symmetric self play for automatic gold discovery in robotic manipulation yet with authors openeye mateas plappert at all. The idea here is to solve a gold conditioned a reinforcement learning. So why, fundamentally, why do you think that splitting the learning problem in his way, using two separate agents,Why is that somehow better?
Play episode from 18:20
Transcript
