肌电和足压信息融合的外骨骼步态识别
Research on exoskeleton gait recognition based on sEMG and foot pressure information introspection
汪步云 1缪龙 2吴臣 2杨鸥 2张振 2许德章1
作者信息
- 1. 安徽工程大学人工智能学院,安徽芜湖 241000;安徽工程大学机械工程学院,安徽芜湖 241000;芜湖云擎机器人科技有限公司,安徽芜湖 241007
- 2. 安徽工程大学机械工程学院,安徽芜湖 241000
- 折叠
摘要
为解决基于单一信号识别步态相位不够精准的问题,开展了动态交互力激励下的人机协同行走的步态识别研究.设计了肌电和足压信息采集的多模态传感器检测硬件平台;分别对单一信号开展滤波降噪、特征提取与降维等预处理;将表征下肢生理信息的肌电信号与运动信息的足压信号相融合,构建了支持向量机-模糊C均值(support vector machine-fuzzy C-mean algorithm,SVM-FCM)多模信息融合的外骨骼助行步态识别算法;开展了人机协同助行实验,实验结果表明:信息融合后的人机步态相位平均识别率达到82.49%,优于使用单一信号的识别效果,验证了多模信息融合算法识别人机协同步态的有效性.本研究可用于下肢外骨骼机器人运动控制,为人机运动相融奠定基础.
Abstract
Gait introspection is not accurate enough using a single type of signal so that human-robot cooperative walking recognition is carried out under dynamic interaction force excitation.Firstly,multimodal sensors detection platform is designed for sEMG and foot pressure information acquisition.Secondly,the signals are preprocessed by filtering,noise reduction,feature extraction and dimension descending.Thirdly,sEMG signals representing the physiological information of the lower limb are introspected with the foot pressure signals of the motion information,and an exoskeleton gait recognition algorithm is established,which supports multi-mode information fusion including vector machine and fuzzy C-means(Support Vector Machine-Fuzzy C-mean algorithm,SVM-FCM).Finally,the experiment of human-robot cooperation is carried out.The experimental results show that the average recognition rate of human-robot gait phase after information fusion reaches 82.49%,which is better than the average recognition rate using a single type of signal.The effectiveness of multimodal information introspection algorithm are verified that human-robot cooperation gait can be recognized.The research could be used for motion control of lower limb exoskeleton robot,which is the foundation of human-robotcompatible cooperation.
关键词
外骨骼机器人/多模态信息感知/人机步态识别/SVM-FCM融合算法Key words
exoskeleton robot/multimodal information introspection/human-robot gait recognition/SVM-FCM fusion algorithm引用本文复制引用
基金项目
国家自然科学基金(61741101)
安徽省重点研究与开发计划(202004a05020013)
安徽省重点研究与开发计划(202004b11020006)
安徽省属公办普通本科高校领军骨干人才项目()
安徽工程大学创新团队()
出版年
2024
NETL
NSTL
万方数据