GIPos: A Geomagnetism/IMU Fusion Positioning Method to Address Geomagnetic Mismatching

Geomagnetic positioning (GP) stands as a formidable research frontier within the domain of indoor positioning. Various GP methods have recently brought remarkable improvements. Nevertheless, the geomagnetic mismatching leading to suboptimal positioning is still serious. To address this problem, we rethink GP in terms of geomagnetic steering features and gross error suppression. We propose a novel method named GIPos, which consists of two stages. In the fusion stage A, we propose a pedestrian dead reckoning (PDR) heading-based geomagnetic distinguishability improvement method. We first construct geomagnetic candidate sets and extract the geomagnetic steering features. Then, a probability model is designed to describe the relationship between geomagnetic steering features and PDR heading, and sequence dynamic search is used to obtain the optimal position corresponding to the geomagnetic test subsequence in this stage. In the fusion stage B, we propose a robust fusion method with forward constraint to fuse PDR and the result of stage A, further reducing the geomagnetic gross error that still exists at seldom points. The performance of the method was evaluated using the MagPIE open dataset containing different indoor scenes. The experimental results show that our GIPos method can achieve superior positioning accuracy compared to other methods. Besides, pedestrian trajectories show the stability and continuity of positioning.