Track2Act: Predicting Point Tracks from Internet Videos enables Generalizable Robot Manipulation figure
AlphaXiv 中文概览(可滚动查看)