Home // International Journal On Advances in Systems and Measurements, volume 14, numbers 1 and 2, 2021 // View article

Estimation of Floor Reaction Force During Walking Using Frequency Analysis

Authors:
Shohei Hontama
Kyoko Shibata
Yoshio Inoue
Hironobu Satoh

Keywords: Discrete Fourier transform; Gait analysis; Average motion mode function; Floor reaction force.

Abstract:
This research group aims to develop a gait analysis system that can be easily used by individuals with little user burden. This research group has proposed a method to estimate the floor reaction force, which is an important parameter in analyzing walking, from the force balance equation using the acceleration measured by wearable sensors, thus obtaining its usefulness. Previously, we reduced the number of sensors from 15 for the whole body part to 5 selected for user burden, but the estimation accuracy decreased. Therefore, we consider how to reduce the number of sensors without sacrificing accuracy. In the proposed method, the acceleration of each part is discrete Fourier transformed and expressed in the frequency domain, a function representing the relationship of acceleration between each part is derived. This function is called the motion mode function, which is used to derive the acceleration in the frequency domain from the measured part to the unmeasured part. The inverse discrete Fourier transform was used to derive the acceleration in the time domain. As a result, the acceleration was estimated with high accuracy. In addition, in order to make the system more versatile, the acceleration of the unmeasured part is estimated using the average motion mode function, which is the average of the motion mode functions obtained from the acceleration of multiple trials by multiple persons and averaged for each order. The estimated acceleration was then used to estimate the composite floor reaction force in two directions. Correlation coefficients were used to check the accuracy of the estimates and demonstrate the usefulness of the proposed method.

Pages: 92 to 102

Copyright: Copyright (c) to authors, 2021. Used with permission.

Publication date: December 31, 2021

Published in: journal

ISSN: 1942-261x