Kalman Filter-Based High-Accuracy Indoor Positioning With NLoS Error Mitigation and Multi-Motion Model Switching

Abstract

In environments where Global Navigation Satellite System (GNSS) signals are unreliable, Ultra-Wideband (UWB) technology stands out for its high-resolution indoor positioning capabilities. However, its performance degrades significantly under Non-Line-of-Sight (NLoS) conditions due to its high-frequency band, which compromises ranging accuracy. Additionally, accurate motion models are vital for precise vehicular indoor positioning.To address these challenges, this work proposes an advanced Indoor Positioning System (IPS) leveraging the Adaptive Kalman Filter with Improved Gain Adjustment (AKF-IGA) for NLoS error mitigation, and a Strong Tracking Cubature Kalman Filter with State Transfer Matrix Self-Adaptation (STCKF-STMSA) for adaptive vehicular motion modeling. The AKF-IGA dynamically adjusts measurement variance to optimize gain under NLoS conditions, while the STCKF-STMSA enables seamless transitions across motion models in various driving scenarios.The performance of the proposed system is rigorously assessed via simulations and real-world experiments, which demonstrates a marked improvement in positioning accuracy. Notably, in NLoS scenario 1, the system maintained an average x-axis error of 0.14 m and y-axis error of 0.08 m, while in NLoS scenario 2, the system maintained an average x-axis error of 0.16 m and y-axis error of 0.11 m.These results highlight the system's proficiency in improving indoor positioning accuracy and reliability, marking a noteworthy contribution to vehicular technology research.