The application of parameter identification methods with competing systems to model a human interface device

Identification methods are applied to experimental data (hand motion) from a design study on a human interface system. The internal model of the Kalman filter employed for the state estimation implicitly utilizes two competing systems (involving an agonist and an antagonist human muscle actuator). The parameters (parsimonious representation) are obtained by minimizing the sums of the squares of the output error residuals, which are tested for whiteness. This technique is a generalization of a method borrowed from pharmacological studies.