Gait identification for an intelligent prosthetic foot

Abstract

Design of an actively controlled prosthetic foot is an emerging research area in robotics. When there are changes in walking conditions such as terrain or speed, classical control methods might confront difficulties. An intelligent prosthetic foot will adapt more efficiently to those changes if it is equipped with an online learning control algorithm. To design such controller, the first step is to acquire real-time gait information from the amputee to study walking behaviors of the individual. In this paper, we developed a neural network-based gait pattern classifier and a rule-based gait phase detector which will provide gait information in real-time.