Towards Effective LiDAR-Based Mapping with Handheld Rotating Sensors in Complex Environments Juyeon Lee (a), Kanghyeock Choi (b*)
a) Program in Smart City Engineering, Inha University, Incheon 22212, Republic of Korea
b) Department of Geoinformatic Engineering, Inha University, Incheon 22212, Republic of Korea
*cwsurgy[at]inha.ac.kr
Abstract
Mobile LiDAR systems provide 3D mapping in complex environments with varied terrain and structures. To overcome the limited field of view of low-resolution LiDAR sensors, additional mechanical rotation is introduced to increase spatial coverage. However, this approach creates challenging conditions with both sparse point distributions and reduced frame-to-frame overlap, posing significant challenges for conventional SLAM algorithms in maintaining reliable scan matching and mapping consistency. This study identifies optimal SLAM algorithms for these conditions and enhances their performance through systematic evaluation. We collected point cloud data using a manually rotated Velodyne VLP-16 system in diverse environments including structured indoor spaces and open outdoor areas, relying solely on LiDAR data without GPS, IMU, or camera assistance. Our evaluation compares multiple existing SLAM algorithms to identify which methods demonstrate the highest robustness and mapping consistency when processing sparse, low overlap point cloud sequences. Beyond algorithm comparison, we explored adaptive strategies to enhance performance based on the characteristics of rotating sensors and varying environmental conditions. These strategies include data handling and algorithm configuration techniques specifically designed to address the challenges posed by sparse and irregularly sampled point clouds. The objective is to establish optimal integration strategies that facilitate stable LiDAR data alignment and consistent 3D mapping results. This work contributes to the practical deployment of compact rotating LiDAR systems for LiDAR-only 3D spatial modeling in GPS-denied environments.
