Action Chunking Proximal Policy Optimization with Feedback Correction

Sanghyun Hahn, Jonghyun Choi

Action chunking provides temporal abstraction in reinforcement learning by selecting short action sequences instead of individual actions, but many existing approaches face two lim…