Closed-loop multivariable system identification for the characterization of the dynamic arm compliance using continuous force disturbances: a model study

Abstract

This study presents a new multivariable closed-loop identification technique for estimating the dynamic compliance of the multijoint human arm during posture maintenance. The method is designed for the application of continuous force disturbances that facilitate interaction of the limb with the environment. The dynamic compliance of the arm arises from different physiological mechanisms and is important for maintaining stable postures and to suppress disturbances. Estimates can be useful to analyze the ability of the nervous system to adapt the arm compliance to different types of disturbances and environments. The technique is linear and requires no a priori knowledge of the system. Linear system behavior is justified for posture tasks where the hand position deviates slightly from a reference position. Interaction results in a closed-loop configuration of arm and environment. The problem with previous methods is the restriction to open-loop systems. With the current technique, the dynamic arm compliance is separately estimated from the closed-loop. The accuracy of the identification technique is tested by simulations for different values of the dynamic compliance of the arm and environment and for different methodological parameters. It is concluded that the identification technique is accurate, even for short observation periods and severe noise.