Abstract
The paper reports a result of dynamical system identification for a human standing control. The whole-body behavior is approximated by COM-ZMP (the center of mass and the zero moment point) model, where ZMP is manipulated so as to stabilize COM, and a piecewise-linear state feedback control. It is the piecewise-affine system model identification, which is a chicken-and-egg problem since the equation of motion has to be known for the parameter identification, while the parameters are required for choosing the equation of motion by which the state is ruled. K-means method and EM algorithm are applied to this problem. The result supports that COM-ZMP model with a piecewise-linear state feedback control well approximates the whole-body motion in stance. It is useful since it helps to evaluate the responsiveness and stabilization ability of a subject.
