Abstract
We present AutoMerge, a LiDAR data processing framework for assembling alarge number of map segments into a complete map. Traditional large-scale mapmerging methods are fragile to incorrect data associations, and are primarilylimited to working only offline. AutoMerge utilizes multi-perspective fusionand adaptive loop closure detection for accurate data associations, and it usesincremental merging to assemble large maps from individual trajectory segmentsgiven in random order and with no initial estimations. Furthermore, afterassembling the segments, AutoMerge performs fine matching and pose-graphoptimization to globally smooth the merged map. We demonstrate AutoMerge onboth city-scale merging (120km) and campus-scale repeated merging (4.5km x 8).The experiments show that AutoMerge (i) surpasses the second- and third- bestmethods by 14% and 24% recall in segment retrieval, (ii) achieves comparable 3Dmapping accuracy for 120 km large-scale map assembly, (iii) and it is robust totemporally-spaced revisits. To the best of our knowledge, AutoMerge is thefirst mapping approach that can merge hundreds of kilometers of individualsegments without the aid of GPS.