Evaluating and post-training robot policies through physical interaction is costly and difficult to scale, motivating action-conditioned world models as interactive simulators. However, visually plausible predictions can misrepresent contact dynamics, providing unreliable feedback for policy evaluation and learning. We...
Ze-Bin Xing, Yu-Hang Zheng, Yu-Peng Zheng et al.· 0 citations
Contact-rich manipulation requires robots to coordinate not only where to move, but also how contact and force should evolve throughout an interaction. Yet most robot policies represent actions primarily as motion trajectories, using tactile signals as inputs or auxiliary prediction targets without explicitly encoding...
Yu-Hang Zheng, Yujie Zang, Yu-Chen Liu et al.· 0 citations
GIFT (Guided Intermediate Feature Training), an architecture-flexible framework for learning intermediate features that translates these structures into training-time constraints through geometry alignment, affordance prediction, and goal-region reconstruction, establishes learning functionally structured intermediate...
Yu-Peng Zheng, Xiang Li, Song-En Gu et al.· 2 citations
WALA, a framework for learning executable latent actions from both action-labeled demonstrations and action-free videos, achieves strong performance on RoboTwin, sets a new state-of-the-art result on RoboCasa, and improves both policy performance and generalization in real-world manipulation tasks.
World action models (WAMs) improve robot control by modeling how observations evolve, but generating future observations at test time incurs substantial latency. Fast-WAM removes this process for efficiency; however, our matched implementations show lower generalization for Fast-WAM than for future-aware alternatives,...