1 paper · 1 filter
Chenhuan Liu, Yi Xu, Feng Wu +7
Embodied AI, including vision-language-action and world-action models, must operate reliably in the physical world. Yet methods that perform well in simulation can degrade substant…