AI RESEARCH
LEADER: Learning Reliable Local-to-Global Correspondences for LiDAR Relocalization
arXiv CS.CV
•
ArXi:2604.11355v1 Announce Type: new LiDAR relocalization has attracted increasing attention as it can deliver accurate 6-DoF pose estimation in complex 3D environments. Recent learning-based regression methods offer efficient solutions by directly predicting global poses without the need for explicit map storage. However, these methods often struggle in challenging scenes due to their equal treatment of all predicted points, which is vulnerable to noise and outliers.