Indoor Positioning Systems Based on A Modified Matrix-Weighted Fusion Estimator with Multipath and NLOS Mitigation

In this paper, an indoor positioning system is investigated for an unmanned ground vehicle based on inertial measurement unit (IMU) and ultra-wide band (UWB) integration. Measurement outliers of UWB usually exist due to multipath and non-line-of-sight (NLOS) effects. To describe characteristics of stochastic noises well, heavy-tailed noises modeled by Student’s t distributions are considered. By a NLOS identification method, a local modified fusion estimator is designed based on the Kalman filter framework to fuse the measurements of IMU and UWB. A modified matrix-weighted fusion estimator is proposed for high estimation accuracy and robustness. Sufficient conditions are given for the fusion estimation error to be exponentially bounded in mean square. Finally, effectiveness of the proposed fusion estimator is indicated by experimental results.