基于渐进无迹卡尔曼滤波网络的人体肢体运动估计

杨旭升; 王雪儿; 汪鹏君; 张文安

doi:10.16383/j.aas.c220523

基于渐进无迹卡尔曼滤波网络的人体肢体运动估计

doi: 10.16383/j.aas.c220523

杨旭升^{1, 2,},
王雪儿^{1, 2,},
汪鹏君^3,,
张文安^{1, 2,}

1.
浙江工业大学信息工程学院杭州 310023
2.
浙江省嵌入式系统联合重点实验室杭州 310023
3.
温州大学电气与电子工程学院温州 325035

基金项目: 浙江省 “尖兵” “领雁” 研发攻关计划 (2022C03114), 国家自然科学基金 (62173305), 浙江省自然科学基金 (LD21F030002) 资助

详细信息

作者简介:
杨旭升：浙江工业大学信息工程学院副研究员. 主要研究方向为信息融合估计, 人体运动估计和目标定位. E-mail: xsyang@zjut.edu.cn

王雪儿：浙江工业大学信息工程学院硕士研究生. 主要研究方向为人体运动估计, 信息融合估计. E-mail: wangxueer@zjut.edu.cn

汪鹏君：温州大学电气与电子工程学院教授. 主要研究方向为人工智能, 信息安全. E-mail: wangpengjun@wzu.edu.cn

张文安：浙江工业大学信息工程学院教授. 主要研究方向为多源信息融合估计及应用. 本文通信作者. E-mail: wazhang@zjut.edu.cn

计量
- 文章访问数: 650
- HTML全文浏览量: 142
- PDF下载量: 214
- 被引次数: 0
出版历程
- 收稿日期: 2022-06-24
- 录用日期: 2023-01-11
- 网络出版日期: 2023-06-07
- 刊出日期: 2023-08-21

Estimation of Human Limb Motion Based on Progressive Unscented Kalman Filter Network

YANG Xu-Sheng^{1, 2
,},
WANG Xue-Er^{1, 2
,},
WANG Peng-Jun^3
,,
ZHANG Wen-An^{1, 2
,}

1.
College of Information Engineering, Zhejiang University of Technology, Hangzhou 310023
2.
Zhejiang Provincial United Key Laboratory of Embedded Systems, Hangzhou 310023
3.
College of Electrical and Electronic Engineering, Wenzhou University, Wenzhou 325035

Funds: Supported by Zhejiang Province “Jianbing” “Lingyan” Research and Development Project (2022C03114), National Natural Science Foundation of China (62173305), and Natural Science Foundation of Zhejiang Province (LD21F030002)

More Information

Author Bio:
YANG Xu-Sheng　Associate researcher at the College of Information Engineering, Zhejiang University of Technology. His research interest covers information fusion estimation, human motion estimation, and target positioning

WANG Xue-Er　Master student at the College of Information Engineering, Zhejiang University of Technology. Her research interest covers human motion estimation and information fusion estimation

WANG Peng-Jun　Professor at the College of Electrical and Electronic Engineering, Wenzhou University. His research interest covers artificial intelligence and information security

ZHANG Wen-An　Professor at the College of Information Engineering, Zhejiang University of Technology. His research interest covers multi-sensor information fusion estimation and its applications. Corresponding author of this paper

摘要

摘要: 针对基于表面肌电信号 (Surface electromyography, sEMG) 的人体肢体运动估计建模困难的问题, 提出一种渐进无迹卡尔曼滤波网络 (Progressive unscented Kalman filter network, PUKF-net), 来实现降低肢体运动与sEMG量测的建模难度以及提高肢体运动估计精度的目的. 首先, 设计深度神经网络从sEMG数据中学习肢体运动状态与sEMG量测之间的映射关系和噪声统计特性. 其次, 采用渐进量测更新方法对先验状态估计进行修正, 减小运动估计的线性化误差, 提高PUKF-net模型的稳定性. 通过结合深度神经网络和渐进卡尔曼滤波的优势, 使得PUKF-net具有良好的模型适应性和抗噪能力. 最后, 设计基于sEMG的人体肢体运动估计实验, 验证了PUKF-net模型的有效性. 相较于长短期记忆网络 (Long short-term memory, LSTM) 和其他卡尔曼滤波网络, PUKF-net在肢体运动估计中的均方根误差 (Root mean square error, RMSE) 下降了14.9%, 相关系数R²提高了5.1%.
- 卡尔曼滤波网络 /
- 人体肢体运动估计 /
- 表面肌电信号 /
- 渐进无迹卡尔曼滤波
Abstract: To solve the difficult modeling problem of human limb motion estimation based on surface electromyography (sEMG), a progressive unscented Kalman filter network (PUKF-net) is proposed to reduce the difficulty of modeling limb motion and sEMG measurements and improve the accuracy of limb motion estimation. Firstly, a deep neural network is designed to learn the mapping relationship between limb motion states and sEMG measurements and the statistical property of noise from sEMG data. Secondly, a progressive measurement update method is used to correct the priori state estimate for reducing the linearization error of motion estimation and improving the stability of the PUKF-net. By combining the advantages of deep neural network and progressive Kalman filter, the PUKF-net has good model adaptability and anti-noise capability. Finally, a human limb motion estimation experiment based on sEMG is designed to verify the validity of the PUKF-net. Compared with the long short-term memory (LSTM) and other Kalman filter network models, the root mean square error (RMSE) of PUKF-net in limb motion estimation has decreased by 14.9% and the correlation coefficient R² has increased by 5.1%.
- Kalman filter network /
- human limb motion estimation /
- surface electromyography (sEMG) /
- progressive unscented Kalman filter

HTML全文

图 1 PUKF-net结构

Fig. 1 Structure of PUKF-net

下载: 全尺寸图片幻灯片

图 2 $ \text{LSTM}_Q$, $ \text{LSTM}_R$网络结构

Fig. 2 Network structure of $ \text{LSTM}_Q$ and $ \text{LSTM}_R$

下载: 全尺寸图片幻灯片

图 3 $ \text{LSTM}_h$计算流程

Fig. 3 Calculation process including $ \text{LSTM}_h$

下载: 全尺寸图片幻灯片

图 4 实验设计 ((a) 传感器布局; (b) 关节角度坐标; (c) 轨迹规划; (d) Optitrack采集到手腕关节点轨迹)

Fig. 4 Experiment design ((a) Sensor layout; (b) Joint angle coordinates; (c) Trajectory planning; (d) Track of wrist joint collected by Optitrack)

下载: 全尺寸图片幻灯片

图 5 sEMG分析 ((a) 人体大臂肌肉分布; (b) Myo位置肌肉横截面; (c)协同矩阵$ W$; (d) sEMG原始信号)

Fig. 5 sEMG analysis ((a) Muscle distribution of human upper arm; (b) Cross-section of Myo wearing position; (c) Non-negative matrix factorization comatrix $ W$; (d) Original signal of sEMG)

下载: 全尺寸图片幻灯片

图 6 关节角度估计曲线

Fig. 6 Joint angle estimation curve

下载: 全尺寸图片幻灯片

表 1 测试者身体参数

Table 1 Physiological information of subjects

测试者	年龄	身高 (cm)	体重 (kg)	性别
S1	31	155	65	女
S2	24	161	53	女
S3	29	182	85	男
S4	20	177	61	男
S5	25	173	75	男
S6	28	175	65	男
S7	30	160	47	女
S8	25	171	72	男
S9	22	175	70	男
S10	24	162	50	女
S11	32	159	54	女
S12	29	170	78	男

下载: 导出CSV

表 2 LSTM、LSTM-KF、PUKF-net在测试集上的RMSE和$ \text{R}^2 $

Table 2 RMSE and $ \text{R}^2 $ of LSTM, LSTM-KF, PUKF-net

测试者	RMSE			$ \text{R}^2$
测试者	LSTM	LSTM-KF	PUKF-net	LSTM	LSTM-KF	PUKF-net
S1	15.913	12.668	11.940	0.823	0.896	0.906
S2	24.568	18.677	15.473	0.622	0.748	0.829
S3	19.736	16.996	14.044	0.737	0.825	0.872
S4	20.653	13.315	12.668	0.679	0.863	0.876
S5	26.746	20.675	16.448	0.629	0.761	0.824
S6	16.793	13.664	11.588	0.803	0.880	0.905
S7	22.193	17.164	14.187	0.699	0.852	0.868
S8	17.984	15.241	12.294	0.748	0.827	0.880
S9	22.537	18.464	15.624	0.710	0.817	0.861
S10	24.142	18.555	16.165	0.655	0.809	0.848
S11	14.601	11.271	10.545	0.682	0.792	0.844
S12	19.196	16.137	13.044	0.721	0.804	0.865
平均值	20.422	16.069	13.668	0.709	0.823	0.865

下载: 导出CSV

参考文献(34)

[1]	Ding Q C, Han J D, Zhao X G. Continuous estimation of human multi-joint angles from sEMG using a state-space model. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2017, 25(9): 1518-1528 doi: 10.1109/TNSRE.2016.2639527
[2]	丁其川, 熊安斌, 赵新刚, 韩建达. 基于表面肌电的运动意图识别方法研究及应用综述. 自动化学报, 2016, 42(1): 13-25 Ding Qi-Chuan, Xiong An-Bin, Zhao Xin-Gang, Han Jian-Da. A review on researches and applications of sEMG-based motion intent recognition methods. Acta Automatica Sinica, 2016, 42(1): 13-25
[3]	Liu H, Tao J, Lyu P, Tian F. Human-robot cooperative control based on sEMG for the upper limb exoskeleton robot. Robotics and Autonomous Systems, 2020, 125: Article No. 103350 doi: 10.1016/j.robot.2019.103350
[4]	陈玲玲, 李珊珊, 刘作军, 张燕. 基于表面肌电的下肢肌肉功能网络构建及其应用研究. 自动化学报, 2017, 43(3): 407-417 Chen Ling-Ling, Li Shan-Shan, Liu Zuo-Jun, Zhang Yan. Construction of lower limb's functional muscle network and its application based on surface EMG. Acta Automatica Sinica, 2017, 43(3): 407-417
[5]	Pallotti A, Orengo G, Saggio G. Measurements comparison of finger joint angles in hand postures between an sEMG armband and a sensory glove. Biocybernetics and Biomedical Engineering, 2021, 41(2): 605-616 doi: 10.1016/j.bbe.2021.03.003
[6]	Ao D, Song R, Gao J W. Movement performance of human — robot cooperation control based on EMG-driven Hill-type and proportional models for an ankle power-assist exoskeleton robot. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2017, 25(8): 1125-1134 doi: 10.1109/TNSRE.2016.2583464
[7]	He Y, Li F, Li J K, Liu J S, Wu X Y. An sEMG based adaptive method for human-exoskeleton collaboration in variable walking environments. Biomedical Signal Processing and Control, 2022, 74: Article No. 103477 doi: 10.1016/j.bspc.2021.103477
[8]	张鋆豪, 何百岳, 杨旭升, 张文安. 基于可穿戴式惯性传感器的人体运动跟踪方法综述. 自动化学报, 2019, 45(8): 1439-1454 doi: 10.16383/j.aas.c180367 Zhang Jun-Hao, He Bai-Yue, Yang Xu-Sheng, Zhang Wen-An. A review on wearable inertial sensor based human motion tracking. Acta Automatica Sinica, 2019, 45(8): 1439-1454 doi: 10.16383/j.aas.c180367
[9]	Bahador A, Yousefi M, Marashi M, Bahador O. High accurate lightweight deep learning method for gesture recognition based on surface electromyography. Computer Methods and Programs in Biomedicine, 2020, 195: 105643 doi: 10.1016/j.cmpb.2020.105643
[10]	Wang C, Guo W Y, Zhang H, Guo L L, Huang C C, Lin C. sEMG-based continuous estimation of grasp movements by long-short term memory network. Biomedical Signal Processing and Control, 2020, 59: Article No. 101774 doi: 10.1016/j.bspc.2019.101774
[11]	朱煜, 赵江坤, 王逸宁, 郑兵兵. 基于深度学习的人体行为识别算法综述. 自动化学报, 2016, 42(6): 848-857 Zhu Yu, Zhao Jiang-Kun, Wang Yi-Ning, Zheng Bing-Bing. A review of human action recognition based on deep learning. Acta Automatica Sinica, 2016, 42(6): 848-857
[12]	Yang W, Yang D P, Liu Y, Liu H. Decoding simultaneous multi-DOF wrist movements from raw EMG signals using a convolutional neural network. IEEE Transactions on Human-Machine Systems, 2019, 49(5): 411-420 doi: 10.1109/THMS.2019.2925191
[13]	Liu J, Ren Y P, Xu D L, Kang S H, Zhang L Q. EMG-based real-time linear-nonlinear cascade regression decoding of shoulder, elbow, and wrist movements in able-bodied persons and stroke survivors. IEEE Transactions on Biomedical Engineering, 2020, 67(5): 1272-1281 doi: 10.1109/TBME.2019.2935182
[14]	Xu L F, Chen X, Cao S, Zhang X, Chen X. Feasibility study of advanced neural networks applied to sEMG-based force estimation. Sensors, 2018, 18(10): 3226 doi: 10.3390/s18103226
[15]	Lu Y Z, Wang H, Zhou B, Wei C F, Xu S Q. continuous and simultaneous estimation of lower limb multi-joint angles from sEMG signals based on stacked convolutional and LSTM models. Expert Systems with Applications, 2022, 203: Article No. 117340 doi: 10.1016/j.eswa.2022.117340
[16]	Chai Y Y, Liu K P, Li C X, Sun Z B, Jin L, Shi T. A novel method based on long short-term memory network and discrete-time zeroing neural algorithm for upper-limb continuous estimation using sEMG signals. Biomedical Signal Processing and Control, 2021, 67: Article No. 102416 doi: 10.1016/j.bspc.2021.102416
[17]	Chen C H, Lu C X, Wang B, Trigoni N, Markham A. DynaNet: neural Kalman dynamical model for motion estimation and prediction. IEEE Transactions on Neural Networks and Learning Systems, 2021, 32(12): 5479-5491 doi: 10.1109/TNNLS.2021.3112460
[18]	Zhang J, Wu Y N, Jiao S. Research on trajectory tracking algorithm based on LSTM-UKF. In: Proceedings of the 7th IEEE International Conference on Network Intelligence and Digital Content (IC-NIDC). Beijing, China: IEEE, 2021. 61−65
[19]	Jondhale S R, Deshpande R S. Kalman filtering framework-based real time target tracking in wireless sensor networks using generalized regression neural networks. IEEE Sensors Journal, 2019, 19(1): 224-233 doi: 10.1109/JSEN.2018.2873357
[20]	Lim H, Ryu H, Rhudy M B, Lee D, Jang D, Lee C, et al. Deep learning-aided synthetic airspeed estimation of UAVs for analytical redundancy with a temporal convolutional network. IEEE Robotics and Automation Letters, 2021, 7(1): 17-24
[21]	Li J M, Chen C W, Cheng T H. Motion prediction and robust tracking of a dynamic and temporarily-occluded target by an unmanned aerial vehicle. IEEE Transactions on Control Systems Technology, 2020, 29(4): 1623-1635
[22]	Coskun H, Achilles F, DiPietro R, Navab N, Tombari F. Long short-term memory Kalman filters: Recurrent neural estimators for pose regularization. In: Proceedings of the IEEE International Conference on Computer Vision (ICCV). Venice, Italy: IEEE, 2017. 5524−5532
[23]	Bao T Z, Zhao Y H, Zaidi S A R, Xie S Q, Yang PF, Zhang Z Q. A deep Kalman filter network for hand kinematics estimation using sEMG. Pattern Recognition Letters, 2021, 143: 88-94 doi: 10.1016/j.patrec.2021.01.001
[24]	Zhao C, Sun L, Yan Z, Neumann G, Duckett T, Stolkin R. Learning Kalman Network: a deep monocular visual odometry for on-road driving. Robotics and Autonomous Systems, 2019, 121: Article No. 103234 doi: 10.1016/j.robot.2019.07.004
[25]	Revach G, Shlezinger N, van Sloun R J G, Eldar Y C. Kalmannet: Data-driven Kalman filtering. In: Proceedings of the ICASSP 2021 —— 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). Toronto, ON, Canada: IEEE, 2021. 3905−3909
[26]	Yang X S, Zhang W A, Liu A D, Yu L. Linear fusion estimation for range-only target tracking with nonlinear transformation. IEEE Transactions on Industrial Informatics, 2020, 16(10): 6403-6412 doi: 10.1109/TII.2019.2955931
[27]	张文安, 陈国庆, 杨旭升. UHF-RFID环境下的移动机器人定位方法. 控制与决策, 2018, 33(10): 1807-1812 doi: 10.13195/j.kzyjc.2017.0741 Zhang Wen-An, Chen Guo-Qing, Yang Xu-Sheng. Mobile robot localization method in UHF-RFID. Control and Decision, 2018, 33(10): 1807-1812. doi: 10.13195/j.kzyjc.2017.0741
[28]	李自由, 赵新刚, 张弼, 丁其川, 张道辉, 韩建达. 基于表面肌电的意图识别方法在非理想条件下的研究进展. 自动化学报, 2021, 47(5): 15 Li Zi-You, Zhao Xin-Gang, Zhang Bi, Ding Qi-Chuan, Zhang Dao-Hui, Han Jian-Da. Review of sEMG-based motion intent recognition methods in non-ideal conditions. Acta Automatica Sinica, 2021, 47(5): 955-969
[29]	Julier S, Uhlmann J, Durrant-Whyte H F. A new method for the nonlinear transformation of means and covariances in filters and estimators. IEEE Transactions on Automatic Control, 2000, 45(3): 477-482 doi: 10.1109/9.847726
[30]	Yang X S, Zhang W A, Yu L, Shi L. Performance evaluation of distributed linear regression Kalman filtering fusion. IEEE Transactions on Automatic Control, 2021, 66(6): 2889-2896 doi: 10.1109/TAC.2020.3012638
[31]	郑婷婷, 杨旭升, 张文安, 俞立. 一种高斯渐进滤波框架下的目标跟踪方法. 自动化学报, 2018, 44(12): 2250-2258 Zheng Ting-Ting, Yang Xu-Sheng, Zhang Wen-An, Yu Li. A target tracking method in Gaussian progressive filtering framework. Acta Automatica Sinica, 2018, 44(12): 2250-2258
[32]	Yang X S, Zhao C, Chen B. Progressive Gaussian approximation filter with adaptive measurement update. Measurement, 2019, 148: Article No. 106898 doi: 10.1016/j.measurement.2019.106898
[33]	谢平, 李欣欣, 杨春华, 杨芳梅, 陈晓玲, 吴晓光. 基于表面肌电非负矩阵分解与一致性的肌间协同耦合关系研究. 中国生物医学工程学报, 2017, 36(2): 150-157 doi: 10.3969/j.issn.0258-8021.2017.02.004 Xie Ping, Li Xin-Xin, Yang Chun-Hua, Yang Fang-Mei, Chen Xiao-Ling, Wu Xiao-Guang. Research on the intermuscular synergy and coupling analysis based on surface EMG nonnegative matrix factorization-coherence. Chinese Journal of Biomedical Engineering, 2017, 36(2): 150-157 doi: 10.3969/j.issn.0258-8021.2017.02.004
[34]	佟丽娜, 侯增广, 彭亮, 王卫群, 陈翼雄, 谭民. 基于多路sEMG时序分析的人体运动模式识别方法. 自动化学报, 2014, 40(5): 810-821 Tong Li-Na, Hou Zeng-Guang, Peng Liang, Wang Wei-Qun, Chen Yi-Xiong, Tan Min. Multi-channel sEMG time series analysis based human motion recognition method. Acta Automatica Sinica, 2014, 40(5): 810-821