Knowledge-based indoor positioning based on LiDAR aided multiple sensors system for UGVs

Abstract

In this paper, an environment knowledge-based multiple sensors indoor positioning system is designed and tested. The system integrates a LiDAR sensor, an odometer and a light sensor onto a low-cost robot platform. While, a LiDAR point-cloud-based pattern match algorithm - Iterative Closed Point (ICP) is used to estimate the relative change in heading and displacement of the platform. Based on the knowledge of the construction's structure, outdoor weather, and lighting situation, the light sensor offers an efficient parameter to improve indoor position accuracy with a light intensity fingerprint matching algorithm on low computational cost. The estimated heading and position change from LiDAR are eventually fused by Extended Kalman Filter (EKF) with those calculated from the light sensor measurement. The results prove that the spatial structure and the ambient light information in indoor environment as knowledge base can be utilized to estimate and mitigate the accumulated errors and inherent drifts of ICP algorithm. These improvements lead to longer sustainable sub meter-level indoor positioning for UGVs.