1 paper
Ziying Song, Yuchen Liu, Zhuoran Xu +6
Embodied world models predict the visual consequences of candidate actions before execution. However, existing action-conditioned world models often adopt uniformly weighted visual…