![]() |
Associate professor
Supervisor of Doctorate Candidates
Supervisor of Master's Candidates
Collaborative Localization and Navigation of Embodied Intelligent Agent Swarms: Embodied intelligent agents (such as drones, autonomous vehicles, humanoid robots, etc.) are equipped with a variety of sensors (including visual cameras, LiDAR, UWB, IMU, etc.) and often operate in swarms within complex indoor and outdoor environments. Traditional navigation and localization methods relying on satellite infrastructure face issues such as insufficient accuracy, poor reliability, and even failure. Therefore, it becomes crucial to explore how to achieve infrastructure-free self-localization and navigation using the sensor data carried by the agents themselves. Our research group focuses on designing intelligent odometry front-ends by integrating multimodal large models with classical SLAM algorithms, enabling multi-sensor fusion for odometry front-end design and achieving self-pose estimation in non-ideal or degraded scenarios. Additionally, based on a swarm collaborative system architecture, we develop backend algorithms to fuse observations from multiple agents for collaborative pose estimation, thereby enhancing the accuracy and reliability of navigation and localization through cooperative efforts.