Filters








1 Hit in 1.8 sec

XIRL: Cross-embodiment Inverse Reinforcement Learning [article]

Kevin Zakka, Andy Zeng, Pete Florence, Jonathan Tompson, Jeannette Bohg, Debidatta Dwibedi
2021 arXiv   pre-print
Specifically, we present a self-supervised method for Cross-embodiment Inverse Reinforcement Learning (XIRL) that leverages temporal cycle-consistency constraints to learn deep visual embeddings that capture  ...  for training policies with reinforcement learning.  ...  Cross-embodiment Inverse Reinforcement Learning.  ... 
arXiv:2106.03911v3 fatcat:az5ipr6dwvavbevjzcrr2qhrby