Control design principle of a low-cost humanoid system using a genetic algorithm

Abstract

Abstract This paper describes a control method for stable walking of a low-cost humanoid in which a two-stage genetic algorithm is applied to find energy-efficient walking based on the inverted pendulum model and straight knee posture. The first stage is to achieve stable walking and the second to find energy-efficient walking. The simulation results are then applied to a real robot, which realizes stable walking with a smooth motion and low energy consumption. From these results, it is confirmed that this control system is valid for highly energy-efficient biped walking. Keywords: LOW-COST HUMANOIDSTRAIGHT KNEE POSTUREINVERTED PENDULUM MODELGENETIC ALGORITHM.