Fetching the paper…

RACER: Rich Language-Guided Failure Recovery Policies for Imitation Learning · Around