Humanoid whole-body dynamics, multimodal VLA (RT-2/OpenVLA), Sim2Real physics transfer, end-to-end autonomous driving world models, and ISO 26262 functional safety.
End-to-end Vision-Language-Action (VLA) foundation models, high-fidelity MuJoCo/Isaac Gym simulations, Sim2Real domain transfer, and bipedal dynamic locomotion/manipulation.
Multi-camera/LiDAR/Radar spatiotemporal fusion, 3D Occupancy networks, generative world models for trajectory prediction, end-to-end driving models, and ISO 26262 functional safety.