Title :
Co-evolving semi-competitive interactions of sheepdog herding behaviors utilizing a simple rule-based multi agent framework
Author :
Lakshika, Erandi ; Barlow, Michael ; Easton, Adam
Author_Institution :
Sch. of Eng. & IT, Univ. of New South Wales, Canberra, ACT, Australia
Abstract :
Sheepdog herding behaviors demonstrate an interesting form of interactions between two classes of agents - sheep and the dog. The nature of the interactions between sheep and the dog takes a special form of competition which is different to the traditional prey-predator interactions where the success of prey depends on the failure of the predator and vice versa. In consequent, the development of an appropriate objective function to efficiently co-evolve successful sheepdog herding behaviors becomes challenging. This paper presents a framework to efficiently co-evolve sheepdog herding behaviors utilizing the simple rule based agent approach in order to derive high fidelity behavior dynamics and discusses the challenges involved in the process.
Keywords :
knowledge based systems; multi-agent systems; optimisation; prey-predator interactions; rule-based multiagent framework; semicompetitive interactions coevolution; sheepdog herding behaviors; Complexity theory; Dynamics; Linear programming; Robots; Sociology; Space exploration; Statistics;
Conference_Titel :
Artificial Life (ALIFE), 2013 IEEE Symposium on
Conference_Location :
Singapore
DOI :
10.1109/ALIFE.2013.6602435