Boundary Condition Relaxation Method for Online Motion Planning of Biped Robots

Abstract

A fast online legged motion planning method is proposed. It utilizes a general solution of the equation of motion of an approximate dynamical biped model whose mass is concentrated to the center of mass in the analytical form. Physically feasible referential trajectories are planned from only one step command of desired motion by relaxing boundary condition, namely, by admitting some error between the desired state and actually reached one. It potentially creates responsive motion which requires strong instantaneous acceleration by accepting discontinuity of ZMP. The validity of the proposed method is ensured through both simulations and experiments with a small anthropomorphic robot.