Conducted multi-robot SLAM research and field validation, focusing on autonomous exploration in challenging environments such as tunnels. Designed and implemented map merging techniques to integrate maps from multiple robots under communication-constrained conditions. Researched and applied place recognition and loop closure across robots equipped with LiDARs of different resolutions and fields of view (FoV). Successfully deployed and tested the system on real robots in fully realistic environments, ensuring robustness in practical scenarios. Explored deep learning approaches for point cloud-based place recognition, including: 1. Heterogeneous LiDAR place recognition between point clouds with different resolutions and FoVs. 2. Development of a CPU-efficient real-time place recognition model for deployment on resource-limited platforms.

更多

Implemented LiDAR-based SLAM at ports and docks to support vessel berthing and departure operations. Adapted SLAM techniques for maritime environments, enabling reliable mapping at ports and docks for vessel berthing and departure. Enhanced SLAM performance in challenging maritime environments (ports and docks) to support vessel berthing and departure.

更多