Can a robot learn locomotion in the wild? We found that with the right implementation, it can be a walk in the park ◡̈ I'm excited to share we've just released code to train your own, real A1! Check it out at: github.com/ikostrikov/wal…
For training videos, please see: sites.google.com/berkeley.edu/w… Paper: arxiv.org/abs/2208.07860 thanks to my collaborators @ikostrikov & @svlevine!
@smithlaura1028 wow that's amazing, thanks for gathering this data
@smithlaura1028 This sort of learning may be a path to artificial general intelligence if it could be applied to methods of thinking rather than just methods of walking.
@smithlaura1028 @smithlaura1028 thanks for the code release... where do you get an A1?
@smithlaura1028 Can you discuss your computing environment - including how you interface with the A1? I'd like to possibly replicate this with my GO-1 but it would be easier if I understand the hardware setup also Also it sounds like using VR for reward shaping is not advised, right?
@smithlaura1028 This is so cool, Laura! A good researcher is the best saleswoman for A1. Now I wanna buy one :)
@smithlaura1028 @0knaomi Great now the murder robots can track you down even if you escape the megacorp city.