Physics-based models for human gait analysis

Abstract

This chapter deals with fundamental methods as well as current research on physics-based human gait analysis. We present valuable concepts that allow efficient modeling of the kinematics and the dynamics of the human body. The resulting physical model can be included in an optimization-based framework. In this context, we show how forward dynamics optimization can be used to determine the producing forces of gait patterns. To present a current subject of research, we provide a description of a 2D physics-based statistical model for human gait analysis that exploits parameter learning to estimate unobservable joint torques and external forces directly from motion input. The robustness of this algorithm with respect to occluded joint trajectories is shown in a short experiment. Furthermore, we present a method that uses the former techniques for video-based gait analysis by combining them with a nonrigid structure from motion approach. To examine the applicability of this method, a brief evaluation of the performance regarding joint torque and ground reaction force estimation is provided.