Cooperation and sharing of caught prey in competitive continuous coevolution using the predator-prey domain

Computer Science and Information Systems

article

Krisztián Varga¹  Attila Kiss² 

[1] Eotv ¨ os Lor ¨ and University ´ Budapest;Eotv ¨ os Lor ¨ and University department of Information Systems ´ Budapest

关键词: artificial life;    neural networks;    evolution;    coevolution;    cooperation;    arms race;    competition;    predator;    prey;   

DOI  :  10.2298/CSIS201122022V

学科分类：土木及结构工程学

来源: Computer Science and Information Systems

【 摘 要 】

Competition is one of the main driving factors of evolution and can be observed in nature as well as in simulations. Competition can occur between predators and preys causing an arms race, but it can also happen between individuals of the same species. Our simulation uses the predator-prey domain (with carnivores, herbivores and plants,) and continuous (not generation based) neuro-evolution to create a complex environment where both forms of competition arise. The characteristics of the simulation make it hard for a predator to catch prey alone, this creates a dependence on cooperation. The predators can share the caught prey with nearby members, in order to help them work together. We explore how sharing affects the cooperation of the hunters, and compare the effectiveness of one and two predator populations.

【 授权许可】

CC BY-NC-ND