How to Train a Robot Arm - A New Method

In this video, I discuss the paper "Asymmetric self-play for automatic goal discovery in robotic manipulation," which describes an effective reinforcement learning method for training robots in simulation by using self-play (robot vs robot). One robot's goal is to move the objects around in a way that it thinks the other robot won't be able to copy, and then the other robot receives this end state as its goal and it tries to move the objects around to match it.

My 3 favorite things about this method are:
1. You get an auto-generated curriculum for the robot to learn from.
2. You know that each task is solvable since a robot created it.
3. You automatically get a demonstration of each task that can be learned from.

📄 The Paper:

Authors (OpenAI):
Matthias Plappert, Raul Sampedro, Tao Xu, Ilge Akkaya, Vineet Kosaraju, Peter Welinder, Ruben D'Sa, Arthur Petron, Henrique Ponde de Oliveira Pinto, Alex Paino, Hyeonwoo Noh, Lilian Weng, Qiming Yuan, Casey Chu, Wojciech Zaremba

Connect with me:

[useranonymous]

I got like 10 invention ideas from this video: totally blew my mind and you are awesome for doing this— protein computers are on the way from my brain😀

[YouChube]

diamond in the rough, so very very appreciative of your institutive lessons. With more people like you we may just kickstart this slow slow incremental human progress

[hanyhamed]

Amazing explanation, thank you very much! I wish you continue such videos, fantastic work!!

[GuillermoValleCosmos]

This was very clearly explained! Thank you!

[mcaelen]

Thanks for sharing, it was very interesting !

[shvprkatta]

very interesting!..thank you for sharing!