•  
  •  
 

MEASUREMENT ERROR OF SIGNAL PROCESSING METHODS TO ESTIMATE VERTICAL CENTER OF MASS DISPLACEMENT DURING BIPEDAL HOPPING

Abstract

BACKGROUND: Double integration of the vertical ground reaction force (GRF) signal allows the estimation of center of mass (COM) displacement in various movement tasks. However, defining the integral of the vertical GRF signal over an extended time results in compounding error causing signal drift, which prohibits an accurate estimation of COM displacement. In this study, we compared measurement error of different signal processing methods to estimate vertical COM displacement from the vertical GRF during bipedal hopping. METHODS: Fifteen adults (age: 24±6 y; mass: 71.6±14.0 kg; height: 171.4±7.6 cm) completed 5, 10 and 15 consecutive bilateral hop trials on a force platform recording at 1000 Hz while vertical COM position was recorded using a 3D motion analysis system at 200 Hz. Vertical COM displacement was estimated from the vertical GRF in three ways: 1) double integration without additional signal processing, 2) double integration with linear detrending, and 3) double integration with cubic spline detrending. Vertical position from the motion analysis system was then used to compute root-mean-square error (RMSE) of each signal processing method and was compared using a 3 (method) x 3 (hop test) repeated measures ANOVA (p<0.05). RESULTS: The cubic spline detrending method had significantly less cumulative RMSE (p<0.01; 0.018±0.006) than double integration (0.913±0.945) and linear detrending methods (0.075±0.084). Linear detrending for 10 hops (p<0.01; 0.062±0.038) and 15 hops (p=0.002; 0.135±0.119) had significantly greater RMSE compared to cubic spline detrending (10 hops: 0.016±0.006; 15 hops: 0.018±0.006), but not for 5 hops (p=0.057; linear detrending: 0.028±0.017 vs. cubic spline detrending: 0.020±0.006). Double integration had significantly greater RMSE for all hop tests (p<0.01; 5 hops: 0.151±0.126; 10 hops: 0.651±0.353; 15 hops: 1.938±0.919) compared to linear detrending (5 hops: 0.028±0.017; 10 hops: 0.062±0.038; 15 hops: 0.135±0.019) and cubic spline detrending methods (5 hops: 0.020±0.006; 10 hops: 0.016±0.006; 15 hops: 0.018±0.006). CONCLUSION: Cubic spline detrending had the smallest RMSE when compared to the double integration and linear detrending methods. We recommend cubic spline detrending of the vertical GRF signal should be used to minimize measurement error when estimating vertical COM displacement during bipedal hopping.

This document is currently not available here.

Share

COinS