MPA-PDR: Improved PDR Indoor Positioning Method Under Multiple Pedestrian Activities
IEEE Sensors Journal · 01.06.2026 · 26 (11) · 17398-17413
DOI: 10.1109/jsen.2026.3684572 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Sensors Journal, erschienen am 01.06.2026. Verfasst von Shuai Zhang und 4 weiteren Autor:innen. Die Arbeit nennt 44 Literaturangaben. Gefördert von: Science and Technology Research Program of Chongqing Municipal Education Commission, Natural Science Foundation of Chongqing, China, Team Building Project for Graduate Tutors in Chongqing, China.
Abstract (Originalsprache)
With the increase in large indoor scenarios and complex integrated buildings, the demand for indoor positioning has increased. Currently, the Pedestrian Dead Reckoning (PDR) positioning technology based on smartphones has attracted considerable attention due to its lack of the need for additional equipment and without relying on external infrastructure. Variations in smartphone pose and carrying mode under different pedestrian activities can seriously degrade step detection and bias the estimation of step length and heading during PDR positioning, leading to a reduction in positioning performance. To address these issues, this article proposes an improved method for PDR positioning under multiple pedestrian activities (MPA-PDR). In this method, a step detection algorithm considering pedestrian activities is first designed, which uses convolutional neural network (CNN)-support vector machine (SVM) to identify common pedestrian activities, and different time thresholds of step period and peak thresholds are set in the step detection process for different pedestrian activities to enhance the accuracy of multiactivity step detection. Then, an adaptive step length estimation (SLE) model [Transformer–long short-term memory (LSTM)] is proposed, which can obtain the appropriate step length based on different pedestrian activities and enhance the accuracy of SLE. Finally, the heading angle is corrected for the problem of inconsistent movement direction of pedestrians and the smartphone orientation under different pedestrian activities. Experimental results show that the proposed method reduces the mean positioning error by 28% in the walking-flat scene, 11% in the walking-phone scene, 16% in the running-flat scene, and 21% in the mixed activity scene and significantly improves the positioning effect, indicating that the improved method has good adaptability for multiactivity positioning and provides certain possibilities for large-scale intelligent indoor positioning.
Förderung
- Science and Technology Research Program of Chongqing Municipal Education Commission
- Natural Science Foundation of Chongqing, China
- Team Building Project for Graduate Tutors in Chongqing, China
Zitieren
APA
Zhang, S., Zhou, E., Li, Z., Chen, R., & Yan, X. (2026). MPA-PDR: Improved PDR Indoor Positioning Method Under Multiple Pedestrian Activities. IEEE Sensors Journal, 26(11), 17398-17413. https://doi.org/10.1109/jsen.2026.3684572
Vancouver
Zhang S, Zhou E, Li Z, Chen R, Yan X. MPA-PDR: Improved PDR Indoor Positioning Method Under Multiple Pedestrian Activities. IEEE Sensors Journal. 2026;26(11):17398-17413. doi:10.1109/jsen.2026.3684572
BibTeX
@article{Zhang2026,
author = {Zhang, Shuai and Zhou, Enjie and Li, Zhaoyang and Chen, Ruizhi and Yan, Xiaofeng},
title = {MPA-PDR: Improved PDR Indoor Positioning Method Under Multiple Pedestrian Activities},
journal = {IEEE Sensors Journal},
year = {2026},
volume = {26},
number = {11},
pages = {17398--17413},
doi = {10.1109/jsen.2026.3684572}
}
EUVIMED