Who We Are

We study intelligent systems that tightly connect perception, reasoning, and action. Our research lies at the intersection of computer vision, multimodal foundation models, and embodied intelligence, with a strong emphasis on building AI systems that can operate reliably in the real world.

Rather than focusing solely on isolated models or benchmarks, we aim to develop robust, scalable, and deployable AI technologies that integrate sensing, decision-making, and control. Our work spans both fundamental AI research and system-level implementations, addressing challenges such as generalization under distribution shift, data efficiency, and real-world constraints.

• Designing generative and multimodal models that enable controllable and trustworthy perception.
• Developing Vision–Language–Action (VLA) frameworks for embodied agents and robotic systems.
• Building end-to-end AI systems that bridge simulation, learning, and physical execution.

If you are interested in pushing the boundaries of real-world AI systems and working on challenging problems at the intersection of theory and practice, we encourage you to contact us and explore opportunities to join the lab.