Robot-DIFT: Distilling Diffusion Features for Geometrically Consistent Visuomotor Control
返回详情
VLA / Vision-Language-Action 每日论文卡