人体动作和路况的快速准确识别是实现智能假肢自主控制的基础与前提。本文提出了一种基于假肢(下肢)惯导信号的高斯混合模型(GMM)和隐马尔可夫模型(HMM)融合的人体动作和路况识别方法。首先,使用惯性传感器采集膝关节处 x、y 和 z 轴方向上的加速度、角度和角速度信号,然后用时间窗截取信号段并用小波包变换消除信号的抖动噪声;接着对预处理后的信号进行快速傅里叶变换,提取其系数作为特征值;随后对特征进行主成分分析(PCA),去除冗余信息;最后采用高斯混合模型和隐马尔可夫模型进行假肢动作和路况识别。试验结果表明,本文方法对常规的动作(散步、跑步、骑行、上坡、下坡、上楼梯和下楼梯)的识别率分别达到 96.25%、92.5%、96.25%、91.25%、93.75%、88.75% 和 90%。同等试验条件下,将本文方法与常规的支持向量机(SVM)识别方法进行比较,结果显示本文方法的识别率明显较高。本文研究结果或可为智能假肢的监测和控制提供新的思路和途径
References
[1]
1. Sup F, Bohara A, Goldfarb M. Design and control of a powered transfemoral prosthesis. Int J Rob Res, 2008, 27(2): 263-273.
[2]
2. Tucker M R, Olivier J, PAGEL Anna, et al. Control strategies for active lower extremity prosthetics and orthotics: a review. J Neuroeng Rehabil, 2015, 12(1): 1-29.
[3]
6. Headon R, Curwen R. Recognizing movements from the ground reaction force, CiteSeer, 2001:1-8.
[4]
7. Jeong J, Cho W, Kim Y, et al. Recognition of lower limb muscle EMG patterns by using neural networks during the postural balance control//3rd Kuala Lumpur International Conference On Biomedical Engineering 2006, Berlin: Springer, 2007: 82-85.
10. Wang Y, Shi Y, Wei G. A novel local feature descriptor based on energy information for human activity recognition. Elsevier Science Publishers B. V, 2017: 19-28.
[7]
12. Jiménez-Fabián R, Verlinden O. Review of control algorithms for robotic ankle systems in lower-limb orthoses, prostheses, and exoskeletons. Med Eng Phys, 2012, 34(4): 397-408.
15. Preece S J, Goulermas J Y, Kenney L P, et al. A comparison of feature extraction methods for the classification of dynamic activities from accelerometer data. IEEE Trans Biomed Eng, 2009, 56(3): 871-879.
[10]
17. Abhayasinghe N, Murray I. Human activity recognition using thigh angle derived from single thigh mounted IMU data//2014 International Conference On Indoor Positioning And Indoor Navigation (IPIN), IEEE, 2014: 111-115.
[11]
18. Shi G Y, Zou Y X, Li W J, et al. Towards multi-classification of human motions using micro IMU and SVM training process. Adv Mat Res, 2009, 60-61: 189-193.
[12]
20. Panahandeh G, Mohammadiha N, Leijon A A. Continuous hidden markov model for pedestrian activity classification and gait analysis. IEEE Trans Instrum Meas, 2013, 62(5, SI): 1073-1083.
[13]
22. Oskoei M A, Hu Huosheng. Myoelectric control systems-A survey. Biomed Signal Process Control, 2007, 2(4): 275-294.
5. Powers C M, Boyd L A, Torburn L, et al. Stair ambulation in persons with transtibial amputation: an analysis of the Seattle LightFoot. J Rehabil Res Dev, 1997, 34(1): 9-18.
13. Kim S K, Hong S, Kim D. A walking motion imitation framework of a humanoid robot by human walking recognition from IMU motion data//IEEE-Ras International Conference on Humanoid Robots, IEEE, 2009:343-348.
[20]
16. Kwapisz J R, Weiss G M, Moore S A. Activity recognition using cell phone accelerometers. Acm Sigkdd Explorati, 2011, 12(2): 74-82.
[21]
19. Dehzangi O, Taherisadr M, Changalvala R. IMU-based gait recognition using convolutional neural networks and multi-sensor fusion. Sensors (Basel), 2017, 17(12): 2735-2757.
[22]
21. You Y, Qian Y, He T, et al. An investigation on DNN-derived bottleneck features for GMM-HMM based robust speech recognition//IEEE China Summit and International Conference on Signal and Information Processing, IEEE, 2015: 30-34.
[23]
23. Zhou X, Zhou C, Stewart B G. Comparisons of discrete wavelet transform, wavelet packet transform and stationary wavelet transform in denoising PD measurement data//Conference Record of the 2006 IEEE International Symposium on Electrical Insulation, IEEE, 2006: 237-240.