arXivJeremy Collins, Namra Patel, Ayush Agarwal, Shitij Govil, Animesh GargTue, Nov 3, 2026, 7:25 AM PST
score 16.9
Robot policy learns actions by predicting 3D motion instead of video
Original: FloMo: 3D Scene Flow as a World Action Model Intermediate for Learning from Human Video
Source: arxiv.org ↗
Writing ELI5 summary…