Visualization and identification of macroscopic dynamics of a human motor control based on the motion measurement

This paper proposes a new scheme to identify the macroscopic dynamics of humans´ motor control based on the COM-ZMP (the center of mass and the zero-moment point) model, which has been vigorously studied and is widely utilized for a design of humanoid robot controllers. Since the model is mathematically well-defined with a small number of system parameters and control parameters, it is easy to identify both those parameters from loci collected through motion measurements by the least-square method. A difficulty is how to collect loci of sufficiently various motions for reliable identification since a human in general unconsciously stabilizes him/herself at any time and hardly shows behaviors in a distance from the point of equilibrium. This problem is solved by visually understanding the macroscopic behavior of the model in theory and set up protocols to cover wide range of the state space with predictions of behaviors to be observed in each area in a phase portrait. By adding preparatory motion in order to accelerate COM to the desired and discarding the portion of that phase, one can obtain valid loci. We examined the proposed method on a standing stabilization control and conducted the identification.