1 paper
Jiakai Huang, Zhongbo Wu, Siyu Xu +5
World Action Models (WAMs) connect visual prediction with robot control, but supplying predictive context often requires expensive future-video generation. Direct policies avoid th…