Mathmod 2018 Extended Abstracts

Potential Field Based Optimization of a Prey-Predator Multi-Agent System

MATHMOD 2018 Extended Abstract Volume​, ARGESIM Report 55 (ISBN 978-3-901608-91-9), p 127-128, DOI: 10.11128/arep.55.a55287


This paper proposes a framework for the optimization of adversarial potential-based prey-predator-like problems. The adversarial potentials are decomposed onto a basis set with different weights. Each weight is individually optimized in a Particle Swarm Optimization-like manner using a Covariance Matrix Adaptation Evolution Strategy (CMA-ES). The proposed results are illustrated via a "Cops & Robbers" scenario.