RLHND: Video Foundation Models as Physically Grounded Hand Trackers for Robot Learning Paper • 2610.09455 • Published 1 day ago • 19