1 paper
Zehua Fan, Junjie He, Wenxuan Song +14
World action models (WAMs) built on video generation backbones are a rising recipe for robot learning, yet remain confined to tabletop manipulation. Mobile manipulation demands sim…