Asymmetric actor critic for image-based robot learning

In three linesOpenAI releases an asymmetric actor-critic method for robot learning: a lightweight actor runs on-robot while a heavy critic trains offline. Approach tested on object manipulation tasks.

Summary generated by Claude — human-verified