INS/GNSS/LiDAR Fusion for Accurate Positioning of UAVs in Complex River Environments

This paper describes the development and evaluation of an INS/GNSS/LiDAR algorithm aimed at maintaining cm-accuracy when an RTK/PPK solution is not available for airborne data collected by a UAV flying in a complex river environment. Two methods are proposed to derive position and velocity from sparse LiDAR point clouds to be utilized in an Error State Kalman Filter: scan-to-scan matching, not relying on global or local maps, and scan-to-map matching, relying on a temporary local map. Both methods use simple point-to-point ICP. Results suggest that the proposed algorithms perform better than a standalone INS/GNSS solution, while achieving performance comparable to fix/float PPK.