System Identification in Motor Control: Time-Varying Techniques

Abstract

System identification techniques have found great applicability in the description of various dynamic aspects of the motor control system under time-invariant conditions. For example, the dynamic relation between small-amplitude angular displacements applied to the human ankle, and the resulting ankle torque at a constant mean torque, is well described by a linear, time-invariant dynamic-stiffness transfer function (Hunter and Kearney, 1982), The time-invariant dynamic-stiffness identification approach has also been successful in characterizing changes in the dynamic stiffness of the human ankle during slow (both large and small) changes in muscle activation levels (Hunter and Kearney, 1984a). However, during limb movement muscle forces and hence limb torques are often changing relatively rapidly, with the result that the mechanical dynamics can no longer be represented using time-invariant system identification techniques. Time-varying techniques are required.