UWB-Foot-SLAM: Bounding Position Error of Foot-mounted Pedestrian INS with Simultaneously Localized UWB Beacons

Abstract

This paper proposes a Simultaneous Localization And Mapping (SLAM) approach, utilizing a combination of foot-mounted Inertial Measurement Units (IMU), foot-mounted Ultra-WideBand (UWB), and environment-deployed UWBs, referred to as the UWB-Foot-SLAM. The proposed approach first leverages a Zero-velocity-UPdaTe (ZUPT)-aided Inertial Navigation System (INS) to map unknown UWB beacons deployed in an environment during navigation and then utilizes the localized beacons to bound position error propagation. An experimental testbed was developed, and we conducted two experiments to validate the performance of the proposed UWB-Foot-SLAM. Experimental results in the first experiment, which involved a pedestrian walking for 3 minutes and deploying two beacons, showed that the UWB beacons' positions estimated by the proposed UWB-Foot-SLAM had displacement errors of 0.28 [m] and 0.22 [m]. In the second experiment, a pedestrian traveled for 25 minutes in a large multi-floor indoor environment and deployed four beacons. The positions of the pedestrian estimated by ZUPT-aided INS had horizontal and vertical Loop-Closure Errors (LCEs) of 11 [m] and 4.5 [m], respectively. When our proposed UWB-Foot-SLAM was used, the LCEs were reduced to 1.49 [m] and 1 [m], respectively, and the covariances associated with the pedestrian position states were bounded after operating for 155 [s].