1 paper · 1 filter
Shenghe Zheng, Wenbo Li, Jiyao Zhang +4
Real-world robot learning is constrained by the cost of collecting experience and evaluating candidate behaviors. Video generation models offer a scalable foundation for visual sim…