Indoor Localization by Kalman Filter based Combining of UWB-Positioning and PDR

Abstract

Ultra-wideband (UWB) ranging-based indoor positioning system (IPS) is suggested as one of the solutions to satisfy the requirement of decimeter level accuracy in indoor environments. However, UWB ranging error caused in non-line-of-sight (NLOS) environments is inevitable, which reduces accuracy of positioning. To tackle this problem, in this paper, an indoor positioning scheme that combines UWB positioning with pedestrian dead reckoning (PDR) is designed and proposed. First, the proposed scheme improves the performance of PDR utilizing UWB positioning in order to achieve the effect of parameter adaptation used in PDR. To this end, step detection and stride length estimation in traditional PDR are substituted with a deep learning-based speed estimation. In addition, heading estimation is improved by calibrating tilt effect of smartphone with the aid of UWB positioning. Then, with the UWB-assisted PDR (U-PDR), we also propose UWB positioning and U-PDR fusion algorithm using Kalman filter (KF). The proposed fusion algorithm complements UWB positioning and U-PDR based positioning, which improves the accuracy of positioning. Experimental results demonstrates that the performance of the proposed algorithm is better than that of UWB positioning or PDR only algorithms.