LiDAR Localization
LiDAR: Light Detection and Ranging
Overview
LiDAR localization estimates position and orientation by matching incoming 3D scans to maps or previously seen places, supporting navigation where GPS is unavailable or unreliable. Geometric methods use scan matching, continuous distance fields, and pose-graph optimization, often fused with IMU or GNSS; learning-based approaches use point-cloud descriptors, scene coordinates, or bird’s-eye-view representations for place recognition and relocalization. Systems are evaluated on datasets such as KITTI, NCLT, Oxford RobotCar, and Boreas, as well as in settings like forests, tunnels, highways, and indoor buildings. A central challenge is maintaining accurate, efficient tracking when geometry is sparse or repetitive, scenes change, or sensors differ, while recognizing and communicating uncertainty rather than committing to ambiguous poses.
Latest papers
42 papers2 papers in the last four weeks, down 67% on the four weeks before. 0.0% of all new papers.