Research Theme
We build world models that understand environments, predict futures, and support planning so agents gain an internal simulation capability.
Core Issues
- Representation learning and latent-space modeling
- Dynamic system modeling and prediction
- Long-horizon planning and decision-making
- Model-based reinforcement learning
Research Content
- Generative world modeling with Transformers and Diffusion models
- Latent dynamics modeling
- Imagination-based planning methods
- Unified frameworks for world models and control
Application Scenarios
- Robot decision-making
- Autonomous systems
- Complex-system modeling