Research Theme
This direction studies how agents perceive, decide, and act in the physical world to build integrated embodied systems from sensing to control.
Core Issues
- Multimodal perception and representation learning across vision, language, and touch
- Long-horizon task learning and generalization
- Combining imitation learning with reinforcement learning
- Sim-to-real transfer
Research Content
- Vision-language-action (VLA) models
- Robot manipulation and skill learning
- Generalist policy learning
- Data-driven robot learning frameworks
Application Scenarios
- Industrial automation
- Service robotics
- Research automation systems