基于Mamba模式感知的步态检测算法及其在PDR中的应用

doi:10.3969/j.issn.0255-8297.2026.04.011

摘要/Abstract

摘要： 针对步态检测算法在复杂运动模式与多变设备携带姿态条件下易产生漏检与误检的问题，本文提出一种融合Mamba神经网络与有限状态机（finite state machine,FSM）的自适应步态检测算法。通过Mamba网络构建分类模型实现对行人运动模式及手机携带姿态的联合判别，并融合自适应阈值FSM、角速度回溯机制以及基于重力投影的非刚性耦合去噪策略，实现复杂携带条件下步态事件的稳定触发，进一步提升行人航位推算（pedestrian dead reckoning,PDR）在复杂场景下的定位精度与鲁棒性。实验结果表明，所提算法在多种运动模式与手机携带姿态条件下均表现出较高的步态检测准确性，有效抑制了误检与漏检现象，降低了由步态检测环节引入的PDR定位误差。

关键词: 行人航位推算, Mamba神经网络, 室内行人定位, 自适应阈值, 步态检测

Abstract: To address the issues of missed and false detections in gait detection algorithms under complex motion modes and varying device carrying postures, an adaptive gait detection algorithm integrating a Mamba neural network and a finite state machine(FSM) was proposed. A classification model based on the Mamba network was constructed to jointly identify pedestrian motion modes and smartphone carrying postures. Furthermore, this algorithm fused an adaptive-threshold FSM, an angular velocity backtracking mechanism,and a gravity projection-based non-rigid coupling denoising strategy to achieve the stable triggering of gait events under complex carrying conditions, thereby further improving the positioning accuracy and robustness of pedestrian dead reckoning(PDR) in complex scenarios. Experimental results demonstrate that the proposed algorithm achieves high gait detection accuracy under various motion modes and smartphone carrying postures,effectively suppresses false and missed detections, and reduces the PDR positioning errors introduced by the gait detection stage.

Key words: pedestrian dead reckoning(PDR), Mamba neural network, indoor pedestrian positioning, adaptive threshold, gait detection

中图分类号:

TP391
P227.9

周子毅, 刘德儿, 钟崇林, 王源, 张涛, 邱永康. 基于Mamba模式感知的步态检测算法及其在PDR中的应用[J]. 应用科学学报, 2026, 44(4): 669-684.

ZHOU Ziyi, LIU Deer, ZHONG Chonglin, WANG Yuan, ZHANG Tao, QIU Yongkang. Mamba-Based Pattern-Aware Gait Detection Algorithm and Application in Pedestrian Dead Reckoning[J]. Journal of Applied Sciences, 2026, 44(4): 669-684.

参考文献

[1] Yu Y, Chen R Z, Chen L, et al. A novel 3-D indoor localization algorithm based on BLE and multiple sensors[J]. IEEE Internet of Things Journal, 2021, 8(11):9359-9372.
[2] 朱建良,王立雅,薄煜明.行人GNSS/PDR组合导航优化估计方法[J].兵工学报, 2023, 44(10):3137-3145.Zhu J L, Wang L Y, Bo Y M. Pedestrian GNSS/PDR integrated navigation system with graph optimization[J]. Acta Armamentarii, 2023, 44(10):3137-3145.(in Chinese)
[3] 马杰,王健,李智.基于改进PDR算法的室内定位方法研究[J].电子测量与仪器学报, 2024, 38(12):211-217.Ma J, Wang J, Li Z. Research on indoor localization method based on improved PDR algorithm[J]. Journal of Electronic Measurement and Instrumentation, 2024, 38(12):211-217.(in Chinese)
[4] Wang B Y, Liu X L, Yu B G, et al. Pedestrian dead reckoning based on motion mode recognition using a smartphone[J]. Sensors, 2018, 18(6):1811.
[5] Zhang R, Mi J, Li J, et al. A continuous PDR and GNSS fusing algorithm for smartphone positioning[J]. Remote Sensing, 2022, 14(20):5171.
[6] 余佩,韩勇强.几种常用步态检测典型算法的设计比较[J].导航定位与授时, 2018, 5(4):21-26.Yu P, Han Y Q. The comparison of several typical gait detection algorithms[J]. Navigation Positioning&Timing, 2018, 5(4):21-26.(in Chinese)
[7] Luo Y R, Guo C, Su J T, et al. Learning-based complex motion patterns recognition for pedestrian dead reckoning[J]. IEEE Sensors Journal, 2021, 21(4):4280-4290.
[8] 刘宇,李汪润,陈燕苹.基于自适应峰值检测的PDR算法研究[J].电子测量技术, 2023, 46(17):37-42.Liu Y, Li W R, Chen Y P. Research on PDR algorithm based on adaptive peak detection[J].Electronic Measurement Technology, 2023, 46(17):37-42.(in Chinese)
[9] 万蓬勃,李学青,汤运启.一种改进的行人航迹推算算法研究[J].电子测量技术, 2024, 47(11):69-77.Wan P B, Li X Q, Tang Y Q. Research on an improved pedestrian dead reckoning algorithm[J]. Electronic Measurement Technology, 2024, 47(11):69-77.(in Chinese)
[10] 李岩,施忠臣,侯燕青,等.行人惯性定位新动态:基于神经网络的方法、性能与展望[J].自动化学报, 2025, 51(2):271-286.Li Y, Shi Z C, Hou Y Q, et al. Emerging trends in pedestrian inertial positioning:neural network-based methods, performance and prospects[J]. Acta Automatica Sinica, 2025, 51(2):271-286.(in Chinese)
[11] Klein I, Asraf O. StepNet-deep learning approaches for step length estimation[J]. IEEE Access,2020, 8:85706-85713.
[12] Li Y, Zeng G P, Wang L P, et al. Accurate stride-length estimation based on LT-StrideNet for pedestrian dead reckoning using a shank-mounted sensor[J]. Micromachines, 2023, 14(6):1170.
[13] Lin Y R, Yu K G, Zhu F Y, et al. A deep-learning-based monocular VO/PDR integrated indoor localization algorithm using smartphone[J]. IEEE Internet of Things Journal, 2024, 11(22):37312-37323.
[14] Miao Q G, Jia L X, Xie K, et al. A comprehensive survey and taxonomy of Mamba:applications,challenges, and future directions[J]. Information Fusion, 2026, 130:104094.
[15] Zagorski P, Dziwinski T, Tutaj A. Steepest descent quaternion attitude estimator[J]. Aerospace Science and Technology, 2018, 77:1-10.
[16] Hou X Y, Bergmann J. A pedestrian dead reckoning method for head-mounted sensors[J].Sensors, 2020, 20(21):6349.
[17] Young F, Mason R, Wall C, et al. Examination of a foot mounted IMU-based methodology for a running gait assessment[J]. Frontiers in Sports and Active Living, 2022, 4:956889.
[18] 毕京学,甄杰,姚国标,等.面向智能手机的改进有限状态机步态探测算法[J].武汉大学学报(信息科学版), 2023, 48(2):232-238.Bi J X, Zhen J, Yao G B, et al. Improved finite state machine step detection algorithm for smartphone[J]. Geomatics and Information Science of Wuhan University, 2023, 48(2):232-238.(in Chinese)
[19] 李鹏,张治胜,王珂,等.融合自适应步长与地图匹配的智能手机PDR室内定位算法[J].导航定位与授时, 2024, 11(4):128-137.Li P, Zhang Z S, Wang K, et al. PDR indoor positioning algorithm for smartphones integrated adaptive step size and map matching[J]. Navigation Positioning and Timing, 2024, 11(4):128-137.(in Chinese)
[20] 李月香,刘燕,袁涛,等.基于加速度信号的走路模式多级分类算法[J].电子学报, 2009, 37(8):1794-1798.Li Y X, Liu Y, Yuan T, et al. Multiple classifier based walking pattern recognizing algorithm using acceleration signals[J]. Acta Electronica Sinica, 2009, 37(8):1794-1798.(in Chinese)
[21] Van Oeveren B T, De Ruiter C J, Beek P J, et al. Optimal stride frequencies in running at different speeds[J]. PLoS One, 2017, 12(10):e0184273.