Dynamic identification of a human-exoskeleton system

Abstract

Body segment inertial parameters of an exoskeleton and human being are of crucial importance for the development of model-based controllers and for the monitoring of rehabilitation processes. These parameters are usually provided using Computer Aided Design (CAD) models or averaged Anthropometric Tables (AT) for human. However, CAD models are often not sufficiently accurate and particularly if the robot is interacting with the environment, also the use of AT for inertial parameters estimation concerns only variations within a relatively small category of subjects. Thus, robust and simple to use identification methods based on kinematic and dynamometric measurements should be developed. In this paper, an identification model of both the human locomotor apparatus and an exoskeleton system is proposed. The proposed approach is able to estimate accurately and with physical consistency the individual mass, center of mass and inertia tensor elements of each segment of both models. The method is validated in simulation with the Exoskeletal Robotic Orthotics for Walking Assistance (E-ROWA) exoskeleton.