zero-shot generalization

Making In-Context Robot Learning Accessible: A Simple Recipe for Manipulation Tasks

A new study clarifies what robots should learn from video demonstrations by defining four semantic targets: ac…