Assessing manual lifting tasks based on segment angle interpolations

Abstract

This study investigates the effects of the number of interpolation points on the prediction accuracy of segment angle trajectory during lifting. Ten participants performed various lifting tasks while a motion tracking system recorded their movements. Two-point through ten-point equal time-spaced segment angles extracted from major segment trajectory data captured by the motion tracking system were used to re-generate the whole body lifting motion by using polynomial and cubic spline interpolation methods. The root mean square error (RMSE) between the reference (motion tracking system) and the estimated (interpolation method) segment angle trajectories were calculated to quantify the prediction accuracy. The results showed that the cubic spline interpolation will yield a smaller RMSE value than one based on the polynomial interpolation. While increasing the number of interpolation points can reduce the RMSE of the estimated segment angle trajectories, there was a diminishing advantage in continuing to add interpolation points. A sensitivity analysis suggests that if the estimation of the segment angles at each interpolation point deviates considerably from the real value, and cannot be controlled at a low level (<10 (), the use of higher number of interpolation points will not improve the estimation accuracy.