A New Point-Cloud-Based LiDAR/IMU Localization Method with Uncertainty Evaluation

Abstract

This paper describes the design, analysis, and experimental evaluation of a new spherical-grid-based (SGB) localization algorithm. This method combines a light detection and ranging (LiDAR)'s spherically-parametrized point cloud with measurements from an inertial measurement units (IMU) to estimate the position and orientation of a moving vehicle. It also quantifies navigation uncertainty. This grid-based method does not require feature extraction and data association, which are necessary steps in landmark-based localization. In addition, we developed an automated testbed to analyze the probabilistic performance of a landmark-based method and of the new spherical gridbased algorithm. The sample and analytical error distributions for both methods are evaluated in a lab environment.