Adaptive Sequence Hybrid Filter-Based Geomagnetic Navigation Method for Autonomous Vehicles

Geomagnetic navigation is a geomagnetic field-based localization and navigation method that has been widely adopted across various domains. The current algorithms are capable of acquiring absolute positioning data, thereby correcting the cumulative errors inherent in odometry and inertial navigation systems. However, the positioning accuracy of the current algorithms are significantly affected by noise and trajectory shape of odometry, while its substantial computational load results in limited real-time performance. To address these limitations, this paper proposes a geomagnetic navigation method based on adaptive sequence hybrid filter. Firstly, to reduce the impact of measurement errors, we incorporate three additional geomagnetic components into the positioning process to improve its accuracy. Secondly, we propose a matching method based on hybrid filter to rectify trajectory shape error. Finally, to address the low real-time performance of sequence matching, we perform feature selection to eliminate heavily disturbed components and reduce subsequent computational load. Additionally, an adaptive resampling strategy based on Kullback–Leibler distance is employed to further improve positioning speed. Experimental results demonstrate that compared to odometry, the proposed algorithm reduces the positioning error from 11.766m to 2.353m, while achieving an average positioning time of 0.044s.