FORMULATION OF DESIRED ZERO MOMENT POINT TRAJECTORY USING STATISTICAL METHOD

This article explains how to construct desired Zero Moment Point (ZMP) region by statistical method. Gait samples, generated by humanoid soccer robot named Robo-Erectus Senior, are converted to unified ones with standard step length. They are evaluated by ZMP margin and energy consumption. The estimated values are also used as weights in the calculation of means and covariance. Thereby, ZMP quality in term of dynamical stability and energy efficiency can be evaluated by covariance. The less the covariance is, the better the ZMP is. For given covariance, desired ZMP region can be constructed easily and quickly for further optimization and learning.