Optimization of Central Pattern Generator based Quadruped Animation using Time Series with Genetic Algorithm

Abstract

Generating believable quadruped motion is a challenging task for an application like Games, Virtual Reality, and Augmented Reality, where runtime user interaction is needed. In order to successfully generate the believable motion, the runtime or dynamic adjustment to motion gaits is essential. As the use of the Central Pattern Generator (CPG) is vital in generating a range of dynamic quadruped motions, the optimization is crucial for smooth motion curves. We use the Genetic Algorithm (GA) based tuning technique to optimize the motion curves generated through CPG. The quadruped motion generated through CPG is applied on Skeletal joints and then optimized through fitting the artificial motion parameters tuned using a genetic algorithm. The results generated show much smooth and stable quadruped motion with believable gait patterns.