2025 CORL CoRL 2025

GraspVLA: a Grasping Foundation Model Pre-trained on Billion-scale Synthetic Action Data