Flat-Upstairs Gait Switching of Lower Limb Prosthesis via Gaussian Process and Improved Kalman Filter

Abstract

The gait switching ability is one of the important indicators for the practical application of prosthetic limbs, but there is little research on it. This paper proposes a prosthetic gait switching strategy based on phase estimation for smooth gait transition from flat to upstairs. Firstly, the current gait category was identified by the hip Angle features (median and amplitude), and the gait transition from plane to upstairs was carried out based on the smooth change of gait. Later, with the Improved Kalman filter (IKF) to estimate gait phase, the current state’s position in the whole gait can be obtained. Then input phase into gaussian process (GP) under different gait function prediction of knee joint angle, and combined with the gait feature to get the current need of knee joint angle, so the prosthesis can be controlled. This paper compared the predicted trajectory with the collected data and found that this method could predict the knee joint angle good in the process of gait change. Finally, by comparing the phase prediction trajectory before and after filtering, this paper finds that the phase prediction trajectory after filtering has better effect and smaller error.