Iterative learning control for biped walking

In this paper, an iterative learning control (ILC) approach is proposed for biped walking control. The biped robot is powered by applying an impulsive push along the stance leg just before the heel strikes. The ILC law is designed based on Poincaré map, and applied to learn the desired impulsive push at every step in the presence of system uncertainties. The convergence of the proposed ILC approach for biped walking control is proofed, and the simulation results show that the proposed ILC approach for biped walking control can track the desired step length effectively, even if the mass of foot can not be ignored compared to that of pelvis.