Precise Indoor Positioning System for Mobile Robots via Smoothed UWB/IMU Sensor Fusion

Abstract

Indoor positioning systems (IPSs) are the foundations for all indoor location-based services and applications. In this article, we present a precise and robust IPS using ultra wide-band (UWB) technology fused with an inertial measurement unit (IMU). Both technologies are integrated to account for the non-line-of-sight (NLOS) problems arising in a dense challenging environment found within an industrial laboratory in Finland. Besides the conventional estimation techniques e.g. extended Kalman filter (EKF), we employ the Rauch-Tung-Striebel (RTS) smoothing algorithm in addition to a multivariate regression-based offset compensation method to improve the overall positioning accuracy of the system. The recommended number of distributed UWB anchors versus the coverage area is also discussed and tested in this article. The experiments were held by a patrolling mobile robot with millimeter accuracy, which acted as a ground truth reference to all used algorithms. The positioning estimation results showed a superior performance by the proposed method (UWB/IMU EKF-RTS-LR) with mean accuracy of 4.7 cm, and 9.6 cm for more than 95% of the time.