GEAR-VLA: Learning Geometry-Aware Action Representations for Generalizable Robotic Manipulation figure
AlphaXiv 中文概览(可滚动查看)