Two-stage fusion localization based on UWB/PDR/Geomagnetism in underground space

A two-stage fusion positioning model based on UWB/PDR/Geomagnetism is proposed for pedestrian positioning in underground space. Firstly, an improved particle filter (PF) based on regional constraint is applied to PDR/Geomagnetism combination positioning, where the PDR positioning results are used to constrain the geomagnetic matching region. The proposed algorithm improves the inherent blindness of scattered particles of traditional PF, thus to enhance the positioning accuracy. Furtherly, Factor graph (FG) is used to fuse the output of PF above with the UWB positioning results, which effectively overcomes the serious problem of positioning hopping points caused by signal occlusion in some areas of UWB system. Experimental results show that the improved PF can outperform extended Kalman filter (EKF) for PDR/Geomagnetism combination positioning and FG algorithm can provide a higher positioning accuracy for UWB/PDR/Geomagnetism fusion positioning.