基于惯性传感器的步态特征估算方法

Gait Feature Estimation Method Based on Inertial Sensor

近年来,惯性测量单元因其便捷性被广泛应用于步态分析的研究中。针对当前步态估算算法过程中精度低以及适用性不高的问题,提出一种基于惯性传感器的步态特征估算方法。将2个IMU固定在受试者的踝关节上方,对采集到的数据进行坐标变换转移至全局坐标系下并进行步态事件的识别;对线性加速度进行二重积分来获得空间位姿信息,并根据步态周期特征引入一种零速判别方式以及误差补偿方式来减小积分漂移,从中提取出相应的步态特征;以光学动作捕捉系统采集的数据作为黄金标准,将所提方法与其进行对比,其中健康步态与模拟异常步态的跨步长平均测量精度(±标准差)分别为-0.035±0.023 m和-0.040±0.024 m,所估算结果与步态标准之间的皮尔森相关系数均大于0.9,模拟异常步态验证了所提方法在不同人群中的适应性。研究结果表明:所提方法能够较好的实现踝关节相关步态特征的估算,且测量精度在正常步态与模拟异常步态之间没有明显的差异性,提供了一种便于使用的光学运动捕捉替代方法。

In recent years,the inertial measurement units(IMUs)have been widely utilized in gait analysis and research due to their convenience.To address the issues of low accuracy and limited applicability in the current gait estimation algorithm,a gait feature estimation method based on inertial sensors is proposed.Two IMUs are fixed above the ankles of a subject,the collected data are transformed into the global coordinate system,and the gait events are identified.Double integration of linear acceleration is performed to obtain spatial pose information,and a zero-speed discrimination method and an error compensation method tailored to the gait cycle characteristics are introduced to mitigate integral drift,enabling extraction of corresponding gait variables.The data collected by optical motion capture system is taken as the gold standard,and the proposed method is compared with it.The average measurement accuracy(±standard deviation)for stride lengths in healthy gait and simulated abnormal gait is-0.035±0.023 m and-0.022±0.020 m,respectively.Pearson correlation coefficients between the estimated results and the gait standard are all greater than 0.9,and the simulation of abnormal gait demonstrates the adaptability of the proposed method in different populations.The research findings indicate that the proposed method excels in estimating the ankle joint-related gait parameters,with consistent measurement accuracy observed between normal and simulated abnormal gaits,offering a user-friendly alternative optical motion capture method.