Locomotion gait optimization for a quadruped robot

Abstract

This article describes the development of a gait optimization system that allows a fast but stable robot quadruped crawl gait. We focus in the development of a quadruped robot walking gait locomotion that combine bio-inspired Central Patterns Generators (CPGs) and Genetic Algorithms (GA). The CPGs are modelled as autonomous differential equations, that generate the necessary limb movement to perform the walking gait, and the Genetic Algorithm perform the search of the CPGs parameters. This approach allows to explicitly specify parameters such as amplitude, offset and frequency of movement and to smoothly modulate the generated trajectories according to changes in these parameters. It is therefore easy to combine the CPG with an optimization method. A genetic algorithm determines the best set of parameters that generates the limbs movements. We intend to obtain a walking gait locomotion that minimizes the vibration and maximizes the wide stability margin and the forward velocity. The experimental results, performed on a simulated Aibo robot, demonstrated that our approach allows low vibration with a high velocity and wide stability margin for a quadruped walking gait locomotion.