research

Vision-language-action models and autonomous driving

I am interested in learning systems that connect visual perception, language understanding, and action for autonomous agents, with a focus on robust operation in complex driving environments.

Robust multimodal and cooperative perception

My research develops 3D perception systems that learn from cooperative viewpoints and adapt across weather, lighting, and sensor domains. Current directions include unsupervised multi-agent supervision, all-weather LiDAR-radar detection, and aligned day-night driving data.

Efficient federated learning

I study communication- and parameter-efficient federated learning for edge intelligence, including dynamic sparse training, parameter freezing, and spatio-temporal task alternation for multi-task models.

Digital twins for medical devices

My earlier work developed digital-twin and model-based methods for personalized therapy, parameter personalization, and safety analysis of implantable medical devices.