G^3VLA: Geometric inductive bias for Vision-Language-Action Models Paper • 2606.24472 • Published 30 days ago • 3