Feature Optimization for Gait Phase Estimation with a Genetic Algorithm and Bayesian Optimization

For gait phase estimation, time-series data of lower-limb motion can be segmented according to time windows. Time-domain features then calculated from the signal enclosed in a window. A set time-domain is used for estimation. In this approach, components feature and length window are influential parameters However, optimal parameter values, which determine its vary across subjects. Previously, these were determined empirically, led degraded estimation performance. To address problem, paper proposes new extraction approach. Specifically, selected using binary genetic algorithm, through Bayesian optimization. two optimization techniques integrated conduct dual task. The proposed method validated five walking running motions. walking, approach reduced error 1.284% 0.910%, while running, decreased 1.997% 1.484%.