1 paper
GigaBrain Team, Boyuan Wang, Bohan Li +23
Vision-language-action (VLA) models that directly predict multi-step action chunks from current observations face inherent limitations due to constrained scene understanding and we…