OC-VLA++: Monocular Geometry-Guided Cross-View Consistency for Viewpoint-Robust Robotic Manipulation figure
AlphaXiv 中文概览(可滚动查看)