CPT-SFM: Conical-planar target-based stepwise feature matching calibration for heterogeneous multi-LiDAR systems

A primary challenge in LiDAR-based autonomous navigation for robotic platforms is the precise calibration of multiple LiDAR sensors to minimize blind areas. This study proposes a novel calibration target, the conical-planar target (CPT), and a calibration method called stepwise feature matching (SFM). This approach is designed to integrate heterogeneous multi-LiDAR sensors into a unified coordinate system. The CPT, combining conical and planar structures, provides stable geometric features irrespective of the LiDAR sensor position and orientation, allowing precise calibration using only a single stationary target without repositioning. The SFM method reduces scan noise through planarization and robustly extracts feature points by registering with a target model. By integrating the feature-point-based initial alignment with scan matching, the proposed approach achieves precise calibration between heterogeneous LiDAR sensors differing in resolution, fields of view, and scanning mechanisms. This study demonstrates the necessity of precise calibration techniques through a quantitative analysis of how decreased calibration accuracy affects the geometric integrity of spatial data. Furthermore, the proposed method achieves high-precision calibration across various sensor combinations, even with minimal sensor overlap and a single calibration target, while consistently extracting feature points across various CPT positions and orientations. Finally, simultaneous localization and mapping experiments using the calibrated multi-LiDAR data validate the practical applicability of the proposed method in autonomous robotic systems.