Living Emerging Worlds for Games

Abstract

In this paper we propose an agent-based, artificial-life inspired, model for living emergent worlds intended to provide realistic behaviors to, already graphically sophisticated, game environments. The ultimate goal of such a proposal is to create the impression to play in a real, living environment. The model is conceived as a physical world where several layers of geographic characteristics and sources of influences define the environment for the living entities. These former are agents that react, based on very simple rules, to the state of the environment and to the influences of other agents. The combined behavior of all these agents produces a dynamic environment that changes gently, but constantly in an unpredictable, although plausible, manner. The proof of the concept is established through the simulation of such a world performed on a fully-functional implementation of the model. We present an example of a simulation and discuss the encouraging results of this and other simulations. These results show that it is possible to create complex and stable ecosystems using simple sets of rules. However, it appears also clear that setting some of the parameters to obtain an acceptable behavior is a delicate task that deserves further exploration and development.