Generative imitation learning policies excel at trajectory planning but lack closed-loop force regulation, while directly incorporating force modalities often requires redesigning or retraining the network. We present UniReflex, a universal plug-and-play framework that equips frozen generative policies with variable im...
Yan Huang, Shou-Jie Li, Zi-Wu Song et al.· 2 citations· ⚡1
Learning contact-rich dexterous manipulation from human demonstrations provides a scalable source of interaction data, yet transferring such skills to robots remains challenging due to unreliable human--robot correspondence. Existing human-to-robot transfer methods typically rely on visual appearance or motion similari...
Xing-Ting Li, Yi-Fan Han, Zi-Jian Lin et al.· 0 citations
PACT-WAM is introduced, a world-action model that jointly generates a 16-step action trajectory and its temporally corresponding visual forecast through conditional flow sampling, andlations show that hierarchical history allocation and joint action-visual generation improve control success.
Yu-Shan Liu, Jing-Jing Fan, Shou-Jie Li et al.· 3 citations
CheckVLA is proposed, which verifies execution with a separately trained, frozen action-conditioned world model, and results support action-conditioned verification as a way to restore feedback during chunked execution while keeping the repair consistent with inference latency.
