RT-H: Action Hierarchies Using Language

Suneel Belkhale, Tianli Ding, Ted Xiao +6

Language provides a way to break down complex concepts into digestible pieces. Recent works in robot imitation learning use language-conditioned policies that predict actions given…