도심환경에서의 변화에 강인한 LiDAR 위치인식 기법

Abstract

This paper presents a LiDAR-based localization method that remains robust to long-term map changes in urban environments. While most existing localization approaches rely on up-to-date high-definition maps, our method achieves accurate pose estimation even when using outdated maps built years earlier. Built upon the KISS-ICP framework, the proposed system introduces adaptive map registration and global consistency maintenance against structural changes such as new buildings, roadworks, and vegetation growth. Experimental results in campus and urban settings demonstrate globally consistent alignment within 7 cm despite multi-year map discrepancies, confirming the method’s effectiveness for long-term autonomous navigation.