An optimized Kalman filter for the estimate of trunk orientation from inertial sensors data during treadmill walking
An optimized Kalman filter for the estimate of trunk orientation from inertial sensors data during treadmill walking
复制标题
DOI：
10.1016/j.gaitpost.2011.08.024
复制
发表时间：
2012-01-01
期刊：
影响因子：
2.4
通讯作者：
Cappozzo, Aurelio
中科院分区：
文献类型：
--
作者：
Mazza, Claudia;Donati, Marco;Cappozzo, Aurelio
本研究的目的是微调的卡尔曼滤波器，目的是提供最佳的估计，在跑步机行走在不同的速度，使用测得的线加速度和角速度分量表示在当地的参考系统在额状面和矢状面的下躯干方向。使用惯性测量单元（IMU）和立体摄影测量系统，同时收集数据，从三个健康的受试者走在跑步机上的自然，缓慢和快速的速度。这些数据被用来估计卡尔曼滤波器的参数，最大限度地减少过滤器提供的躯干方向和通过立体摄影测量获得的方向之间的差异。然后使用优化的参数来处理从另外15名健康受试者收集的数据，这些受试者具有两种性别和不同的人体测量学，执行相同的行走任务，目的是确定滤波器设置的鲁棒性。滤波器被证明是非常强大的。通过IMU和通过立体摄影测量估计的角度之间的差异的均方根值小于1.0度，并且相应曲线之间的相关系数大于0.91。所提出的滤波器设计可以用来可靠地估计躯干横向和正面弯曲行走过程中从惯性传感器数据。需要进一步的研究来确定最适合其他运动任务的滤波器参数。(C)出版社：Elsevier B.V.
The aim of this study was the fine tuning of a Kalman filter with the intent to provide optimal estimates of lower trunk orientation in the frontal and sagittal planes during treadmill walking at different speeds using measured linear acceleration and angular velocity components represented in a local system of reference. Data were simultaneously collected using both an inertial measurement unit (IMU) and a stereophotogrammetric system from three healthy subjects walking on a treadmill at natural, slow and fast speeds. These data were used to estimate the parameters of the Kalman filter that minimized the difference between the trunk orientations provided by the filter and those obtained through stereophotogrammetry. The optimized parameters were then used to process the data collected from a further 15 healthy subjects of both genders and different anthropometry performing the same walking tasks with the aim of determining the robustness of the filter set up. The filter proved to be very robust. The root mean square values of the differences between the angles estimated through the IMU and through stereophotogrammetry were lower than 1.0 degrees and the correlation coefficients between the corresponding curves were greater than 0.91. The proposed filter design can be used to reliably estimate trunk lateral and frontal bending during walking from inertial sensor data. Further studies are needed to determine the filter parameters that are most suitable for other motor tasks. (C) 2011 Published by Elsevier B.V.