基于下肢康复机器人多传感器的异常步态分类研究

Abnormal Gait Classification Based on Multi-sensors in Lower Limb Rehabilitation Robot

针对目前异常步态临床诊断主观性强、效率低、适用于下肢康复机器人的异常步态分类系统紧缺等问题,提出了一种基于下肢康复机器人多传感器的异常步态分类方法。该方法提取了三个特征:左右力曲线相似度、腿部轨迹曲率、步态对称性,将其与异常步态分类标签一起作为KNN模型的输入参数,实现了异常步态分类。设计并实现了下肢康复机器人的行走训练人机交互软件,并将传感信息采集模块集成于其中,实现患者使用机器人行走训练时,软件同步采集各传感器实时信息。该研究招募四名健康受试者进行实验,分别用正常步态、模拟偏瘫步态、模拟帕金森步态进行实验。结果表明该方法可以准确分类出正常步态、左偏瘫步态、右偏瘫步态以及帕金森步态,证明了其有效性。

An abnormal gait classification method based on multiple sensors for lower limb rehabilitation robot is proposed.This method extracts three features:left-right force curve similarity,leg trajectory curvature and gait symmetry,which are used together with abnormal gait classification labels as input parameters of KNN model to realize abnormal gait classification.A human-computer interactive software for walking training of the lower limb rehabilitation robot is designed and implemented.The sensor information acquisition module is integrated into it,so that the software can synchronously collect real-time information of each sensor when patients used the robot for walking training.In this study,four healthy subjects are recruited to carry out experiments with normal gait,simulated hemiplegic gait and simulated Parkinson gait.