基于可穿戴传感器的人体运动识别算法

doi:10.11896/jsjkx.241200083

计算机科学 ›› 2026, Vol. 53 ›› Issue (2): 342-348.doi: 10.11896/jsjkx.241200083

基于可穿戴传感器的人体运动识别算法

蒋磊¹, 王子¹, 杨荣^2,3, 韩旺林¹

1 中国矿业大学(北京)人工智能学院北京 100083
2 国家康复辅具研究中心北京 100176
3 北京市老年功能障碍康复辅助技术重点实验室北京 100176

收稿日期:2024-12-20 修回日期:2025-01-18 发布日期:2026-02-10
通讯作者: 杨荣(yangrong_nrra@163.com)
作者简介:(leijiang@cumtb.edu.cn)
基金资助:
2023年度民政部重点实验室和工程技术研究中心开发课题资助项目

Human Motion Recognition Algorithm Based on Wearable Sensors

JIANG Lei¹, WANG Zi¹, YANG Rong^2,3, HAN Wanglin¹

1 College of Artificial Intelligence,China University of Mining and Technology,Beijing 100083,China
2 National Research Center for Rehabilitation Technical Aid,Beijing 100176,China
3 Beijing Key Laboratory of Rehabilitation Technical Aids for Old-Age Disability,Beijing 100176,China

Received:2024-12-20 Revised:2025-01-18 Online:2026-02-10
About author:JIANG Lei,born in 1982,associate professor,master’s supervisor.His main research interests include artificial intelligence,adaptive control and exo-skeleton robotics.
YANG Rong,born in 1987,senior engineer.Her main research interests include automatic control,brain-computer interface,pattern recognition and signal processing.
Supported by:
2023 Development Program for Key Laboratories and Engineering Technology Research Centers of the Ministry of Civil Affairs.

摘要/Abstract

摘要： 在全球老龄化的背景下,膝关节外骨骼被广泛应用于老年人膝关节的健康维护和康复训练。膝关节外骨骼往往采用嵌入式设备进行人体下肢运动状态的识别,这需要在传感器的选择与布局、算法的准确性与计算复杂度之间找到平衡点。对此,提出了一种适用于膝关节外骨骼的人体运动识别算法。该算法利用大小腿部两个惯性测量单元(IMU)采集下肢运动数据,识别方法包括特征组合、特征筛选和运动状态识别3个步骤。通过交叉方法进行特征组合,以优化特征表达。改进的一对多(One-vs.-Rest,OvR)方法被应用于解决运动识别问题。该方法中使用了一种结合了Relief算法、皮尔森相关系数和机器学习反向筛选的方法的融合算法进行特征选择,以降低计算复杂性,并将历史信息以及其他状态数据整合到模型训练中,以进一步提高准确性。该模型对人体6种日常运动状态进行分类,识别准确率可达97.76% 。实验结果验证了该算法在有限的传感器个数限制下,可以准确且快速地识别下肢运动状态,为膝关节外骨骼的精准检测以及实时控制提供一个准确、低算力要求的解决方案。

关键词: 运动识别, IMU传感器, 特征组合, 机器学习, OvR级联框架

Abstract: Against the backdrop of global aging,knee exoskeletons are widely used in the maintenance of knee joint health and rehabilitation training for the elderly.Knee exoskeletons often employ embedded devices for the recognition of lower limb motion states,which requires finding a balance between the selection and layout of sensors,and the accuracy and computational complexity of algorithms.Therefore,this paper studies a human motion recognition algorithm suitable for knee exoskeletons,which uses two inertial measurement units(IMUs) on the thigh and lower leg to collect lower limb motion data.The recognition method includes three steps:feature combination,feature selection,and motion state recognition.It optimizes feature representation through cross-method feature combination.The improved One-vs.-Rest(OvR) strategy is applied to address motion recognition issues,within which a fusion algorithm combining Relief with Pearson correlation coefficients and a machine learning backward selection method is used for feature selection to reduce computational complexity,and historical information along with other state data are integrated into the model training to further enhance accuracy.The model classifies six types of daily motion states with an accuracy rate of up to 97.76%.Experimental results verify that the proposed algorithm can accurately and quickly recognize lower limb motion states under the limitation of a limited number of sensors,providing an accurate and low computational requirement solution for precise detection and real-time control of knee exoskeletons.

Key words: Motion recognition, IMU sensors, Feature combination, Machine learning, OvR cascade framework

中图分类号:

TP242

蒋磊, 王子, 杨荣, 韩旺林. 基于可穿戴传感器的人体运动识别算法[J]. 计算机科学, 2026, 53(2): 342-348. https://doi.org/10.11896/jsjkx.241200083

JIANG Lei, WANG Zi, YANG Rong, HAN Wanglin. Human Motion Recognition Algorithm Based on Wearable Sensors[J]. Computer Science, 2026, 53(2): 342-348. https://doi.org/10.11896/jsjkx.241200083

参考文献

[1]CHENG X J,HU G Q.Research progress on health impacts of population aging[J].Chinese Journal of Epidemiology,2020,41(11):1915-1920.
[2]KITTISARES S,IDE T,NABAE H,et al.Ergonomic dual four-bar linkage knee exoskeleton for stair ascent assistance[EB/OL].https://doi.org/10.3389/frcrt.2023.1285520.
[3]LIU B,LIU Y W,ZHOU Z,et al.Control of a soft exoskeleton robot for the human knee joint based on dynamic modeling[EB/OL].https://doi.org/10.1017/s0263574722000029.
[4]HUANG Y C,HE,Z X,LIU Y X,et al. Real-time intended knee joint motion prediction by deep-recurrent neural networks[J].IEEE Sensors Journal,2019,19(23):11503-11509.
[5]SONG J Y,ZHU A B,TU Y,et al.Effects of different feature parameters of sEMG on human motion pattern recognition using multilayer perceptrons and LSTM neural networks[J].Applied Sciences,2020,10(10):3358.
[6]GAUTAM A,PANWAR M,BISWAS D,et al.MyoNet:Atransfer-learning-based LRCN for lower limb movement recognition and knee joint angle prediction for remote monitoring of rehabilitation progress from sEMG[J].IEEE Journal of Translational Engineering in Health and Medicine,2020,8:1-10.
[7]PARK S,PARK F C,CHOI J,et al.EEG-based gait state and gait intention recognition using spatio-spectral convolutional neural network[C]//2019 7th International Winter Conference on Brain-Computer Interface(BCI).2019:1-3.
[8]CHAISAEN R,AUTTHASAN P,MINGCHINDA N,et al.Decoding EEG Rhythms During Action Observation,Motor Imagery,and Execution for Standing and Sitting[J].IEEE Sensors Journal,2020,20(22):13776-13786.
[9]ZHOU S,GAO T.Brain Activity Recognition Method Based on Attention-Based RNN Mode[J].Applied Sciences,2021,11:10425.
[10]GOH S K,ABBASS H A,TAN K C,et al.Spatio-spectral representation learning for electroencephalographic gait－pattern classification[J].IEEE Transactions on Neural Systems and Rehabilitation Engineering,2018,26(9):1858-1867.
[11]GORŠIČ M,KAMNIK R,AMBROŽIČ L,et al.Online phase detection using wearable sensors for walking with a robotic prosthesis[J].Sensors,2014,14(2):2776-2794.
[12]AMBROZIC L,GORSIC M,GEEROMS J,et al.CYBERLEGs:A user-oriented robotic transfemoral prosthesis with whole-body awareness control[J].IEEE Robotics & Automation Magazine,2014,21(4):82-93.
[13]YUAN K,WANG Q N,WANG L.Fuzzy-logic-based terrainidentification with multisensor fusion for transtibial amputees[J].IEEE/ASME Transactions on Mechatronics,2014,20(2):618-630.
[14]WANG J,WU D,GAO Y,et al.Integral Real-time Locomotion Mode Recognition Based on GA-CNN for Lower Limb Exoske-leton[J].Journal of Bionic Engineering,2022,19:1359-1373.
[15]HUO W,MOHAMMED S,AMIRAT Y,et al.Fast gait modedetection and assistive torque control of an exoskeletal robotic orthosis for walking assistance[J].IEEE Transactions on Robotics,2018,34(4):1035-1052.
[16]CHEN B J,WANG X G,HUANG Y,et al.A foot-wearable interface for locomotion mode recognition based on discrete contact force distribution[J].Mechatronics,2015,32:12-21.
[17]KONG K,TOMIZUKA M. A gait monitoring system based on air pressure sensors embedded in a shoe[J].IEEE/ASME Transactions on Mechatronics,2009,14(3):358-370.
[18]PARRI A,YUAN K B,MARCONI D,et al.Real-time hybrid locomotion mode recognition for lower limb wearable robots[J].IEEE/ASME Transactions on Mechatronics,2017,22(6):2480-2491.
[19]IGNATOV A.Real-time human activity recognition from acce-lerometer data using Convolutional Neural Networks[J].Applied Soft Computing,2018,62:915-922.
[20]PAYDARFAR A J,PRADO A,AGRAWAL S K.Human Activity Recognition Using Recurrent Neural Network Classifiers on Raw Signals from Insole Piezoresistors[C]//2020 8th IEEE RAS/EMBS International Conference for Biomedical Robotics and Biomechatronics(BioRob).2020:916-921.
[21]WANG C,WU X Y,MA Y,et al.A flexible lower extremityexoskeleton robot with deep locomotion mode identification[J].Complexity,2018,2018:5712108:1-5712108:9.
[22]BIJALWAN V,SEMWAL V B,GUPTA V.Wearable sensor-based pattern mining for human activity recognition:deep lear-ning approach[J].Industrial Robot,2022,49(1):21-33.
[23]HANIF A M,BEQIRI S,KEANE P A,et al.Applications of interpretability in deep learning models for ophthalmology[J].Current Opinion in Ophthalmology,2021,32(5):452-458.
[24]XUE J,CHAN P P,HU X.Experimental study on stacked autoencoder on insufficient training samples[C]//Proceedings of the International Conference on Wavelet Analysis and Pattern Recognition.2017:223-229.
[25]ZHANG Q Y,ZHANG T,ZHAO Z Y.Human motion pattern recognition based on multi-sensor[J].Sensors and Microsystems,2019,38(2):73-76.
[26]ZHAO C.Research on Motion Information Acquisition and Motion Pattern Recognition Technology for Lower Limb Exoskeleton Robot Control[D].Zhengzhou:Zhengzhou University,2020.
[27]FIGUEIREDO J,CAR VALHO S P,GONÇALVES D,et al.Daily Locomotion Recognition and Prediction:A Kinematic Data-Based Machine Learning Approach[J].IEEE Access,2020,8:33250-33262.
[28]BHAKTA K,CAMARGO J,DONOVAN L,et al.MachineLearning Model Comparisons of User Independent & Depen-dent Intent Recognition Systems for Powered Prostheses[J].IEEE Robotics and Automation Letters,2020,5(4):5393-5400.
[29]ZHANG J,GUO J,WANG H,et al.Research on Lower limb Movement Pattern Recognition Method Based on ReliefF-KPCA-SVM[C]//2021 IEEE International Conference on Robotics and Biomimetics(ROBIO).2021:1497-1501.
[30]STOLYAROV R,CARNEY M,HERR H.Accurate HeuristicTerrain Prediction in Powered Lower-Limb Prostheses Using Onboard Sensors[J].IEEE Transactions on Biomedical Engineering,2021,68(2):384-392.
[31]AL-NAIME K,AL-ANBUKY A,MAWSTON G.Human Move-ment Monitoring and Analysis for Prehabilitation Process Management[J].Journal of Sensor and Actuator Networks,2020,9(1):9.
[32]JUAN Y C,LEFORTIER D,CHAPELLE O.Field-aware Factorization Machines in a Real-world Online Advertising System[C]//Proceedings of the 26th International Conference on World Wide Web Companion.2017:680-688.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

基于可穿戴传感器的人体运动识别算法

Human Motion Recognition Algorithm Based on Wearable Sensors

PDF (PC)

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 0

Metrics

本文评价

推荐阅读 0