A robust walking controller optimizing step position and step time that exploit advantages of footed robot

Abstract

This study proposes a robust humanoid step control algorithm that optimizes ground reaction force, step position and step time. Our method is focused on the robot that has finite size of foot and designed to exploit its advantages. The foot allows for the range of ZMP presence and our algorithm use this ZMP range to absorb sensor noise, modeling error and certain amount of disturbances. Thanks to these effect, our step controller is able to produce new stepping time and the stepping position stably. From quadratic programming (QP) technique, we can consider maximum kinematical foot range and maximum foot velocity in the optimization process by setting inequality constraints. The CoM trajectory is re planned in each control cycle with a short cycle preview controller. The linear inverted pendulum model (LIPM) simulation and full dynamics simulation shows that our proposing method precisely reflect the advantages of footed robot and significant improvement in walking robustness to strong perturbation. (C) 2018 Elsevier B.V. All rights reserved.