An indoor 3D localization algorithm integrating barometer–PDR–magnetic fingerprinting

With the increasing demand for indoor navigation, achieving high-accuracy three-dimensional localization in complex multi-story buildings remains a critical challenge. This study proposes a robust particle filtering approach that integrates barometers, inertial measurement units (IMUs), and magnetic fingerprints to improve both the accuracy and robustness of indoor 3D localization. First, for vertical localization, a barometer-based floor recognition algorithm is introduced. This algorithm incorporates walking speed features to refine barometric height differences, thereby improving the accuracy and stability of floor detection. Second, for horizontal localization, short-term trajectory estimates from the IMU are constrained and corrected using magnetic fingerprint matching, which effectively reduces cumulative errors caused by inertial drift. Finally, an improved robust particle filtering algorithm is proposed. It introduces a multi-observation fusion mechanism that incorporates multiple candidate results from magnetic fingerprint matching into particle weight updates. Experimental results show that the proposed method achieves sub-meter accuracy in both horizontal and vertical localization. • To address the low accuracy of conventional floor detection methods in multi-story buildings, this study proposes a barometer-based floor detection algorithm that accounts for variations in floor heights and the dynamic vertical movement of pedestrians on stairs or elevators. • To address the particle degeneracy problem often encountered in PF when integrating PDR with magnetic fingerprints, this study proposes RPF. The method uses multiple magnetic fingerprint matching results as observations to achieve multi-observation fusion, thereby enhancing particle sampling accuracy and improving indoor positioning precision and stability.