Our goal is to build intelligent embodied agents that can perceive, understand, reason, generate, and act in the complex real world. Our research spans embodied AI, multi-modal learning, 3D vision, generative models, robotics, and autonomous driving. Current topics include Vision-Language-Action models, World Action Models, dexterous manipulation, human-robot interaction, humanoid and whole-body intelligence, 3D scene understanding, generative world models, and autonomous driving agents. We aim to bridge foundation models with real robotic systems and make a substantial impact on general-purpose robotics, human-centered AI, and intelligent systems that operate reliably in the physical world.