An Indoor Fusion Positioning Method Based on UWB/IMU/MAG in Harsh Environment

Localization in complex indoor environments remains a serious challenge. To enhance the accuracy of pedestrian position estimation in such an environment, a fused localization framework is proposed. It is primarily driven by ultrawideband (UWB) technology aided by inertial measurement unit (IMU) and geomagnetic (MAG) sensors in nonline-of-sight (NLOS) environments where UWB alone may fail. First, we introduce a MAG fingerprint map lightweighting algorithm based on lower bound functions and matching area constraints, aimed at reducing the number of fingerprints and improving algorithm efficiency. Second, we present an IMU/MAG fusion algorithm utilizing a dynamically weighted k-nearest neighbor (DWKNN), which adjusts the k-value in MAG matching through a method based on fuzzy comprehensive evaluation (FCE). Furthermore, to fully use line-of-sight (LOS) distances when UWB cannot independently estimate location, we propose a UWB/IMU/MAG fusion positioning algorithm based on the tight combination of factor graphs (FGs). Real-world experiments are conducted in a harsh indoor environment, and results demonstrate that our proposed UWB/IMU/MAG fusion framework improves average localization accuracy at least by 71.98% compared to UWB alone and by 18.18% compared to traditional extended Kalman filter (EKF) fusion algorithms. It effectively addresses UWB localization failures in NLOS areas and significantly enhances localization accuracy and robustness in complex environments.