Vid2Robot: End-to-end Video-conditioned Policy Learning with Cross-Attention Transformers | doi.page