1 paper · 2 filters
Yiwei Zhang, Yantai Yang, Jin Gao +7
World Action Models (WAMs) use generative visual representations to condition robot actions, yet how these representations organize action-relevant information remains poorly under…