Divergent Component of Motion-Based Gait Intention Detection Method Using Motion Information From Single Leg

Abstract

This paper proposes a gait intention detection (GID) method based on motion information from a single leg. This method detects three gait states: stair ascending (SA), stair descending (SD), and level walking (LW). The GID is a highly important factor for the control of gait assistive devices. If the GID is limited to the motion information from a single leg, analyzing the gait motions becomes more challenging, owing to the absence of information regarding the phase difference between both legs. To address this challenging, previous GID methods have used fusion data from several sensors. In this paper, only three inertial measurement units are used for proposed single leg GID method. To analyze gait motions using these limited sensors, the divergent component of motion (DCM) of the user is computed. The DCM is a physical quantity incorporating the linear position and linear velocity of the center of mass; these can be computed from the kinematic data of a support leg. In each of the three gait states SA, SD, and LW, the DCM shows different patterns. To classify and detect these different patterns, a pattern recognition method based on an artificial neural network algorithm is used. The GID performance of the proposed method was experimentally evaluated, and it showed 99% success rate for the SA, SD, and LW states.