ICP-based SLAM Using LiDAR Intensity and Near-infrared Data

Abstract

Scan-matching is a method of localization based on the alignment of point clouds measured by sensors such as LiDAR. Most of the scan-matching methods utilize only the shape information of the point cloud. However, depending on the environment, features may not be sufficiently available from the shape information, and the accuracy of alignment may decrease. Therefore, we propose a highly accurate scan-matching method by using sensor fusion to obtain shape and physical features from the environment together. In this paper, the near-infrared information of water puddles, which is often seen inside the damaged nuclear power plant, and the reflection intensity of LiDAR are measured and utilized for localization. Experiment results in the real environment show that the proposed method improved the accuracy of the map and the trajectory of the robot by taking advantage of physical features observed from the environment.