1 paper · 2 filters
Qiao Sun, Liujia Yang, Wei Tang +12
While video-generation-based embodied world models have gained increasing attention, their reliance on large-scale embodied interaction data remains a key bottleneck. The scarcity,…