Research
We study perception, localization, embodied intelligence, and multimodal decision-making for robots, intelligent vehicles, and multi-robot systems.
Robust perception
Point cloud registration and learning-based perception for robots and intelligent vehicles operating in dynamic, large-scale environments.
Localization and state estimation
Reliable localization and estimation methods for autonomous systems, including operation in GPS-denied environments.
Multi-robot systems
Cooperative estimation and autonomous coordination that enable multiple robots to perceive and navigate together.
Embodied intelligence
Learning agents that connect multimodal understanding, spatial reasoning, and action in physical environments.
Vision-language-action models
Multimodal policies that translate visual observations and language instructions into goal-directed robot actions.
Vision-language navigation
Language-guided navigation that combines visual perception, scene understanding, and sequential decision-making.