1 paper
Seungku Kim, Suhyeok Jang, Byungjun Yoon +3
Synthetic data generated by video generative models has shown promise for robot learning as a scalable pipeline, but it often suffers from inconsistent action quality due to imperf…