Real-time Estimation of Lower Extremity Joint Torques in Normal Gait

Abstract

Joint torques of the lower extremity are useful information for gait analysis or gait assistive systems. In this paper, an estimation method for observing the joint torques of the lower extremity is proposed. The proposed method is based on the inverse dynamics of a human body, where the human body model applies a planar link-segment model which consists of seven segments. Lagrangian mechanics is applied to calculate the joint torque, and the required motion information is measured by sensors or estimated by kinematic Kalman filter (KKF). The joint torque equation depends on the ground contact conditions of the human body model, and the ground contact conditions can be identified by gait phases. Thus, multiple set of equations are used for the estimation of the joint torques in different gait phases. The gait phases are detected by Smart Shoes, and the detected gait phases are also used for smoothing the joint torques when the ground contact conditions are changed. The proposed method has been validated by experiments.