GeomVLA: Unifying Scene, Motion, and Action in 3D

Publication
Proceedings of the Conference on Robot Learning (CoRL)