The Gridbrain: an Heterogeneous Network for Open Evolution in 3D Environments

Authors

Abstract

A reasoning system for autonomous agents under open evolution is presented. The system's design goal is to bridge the gap between Artificial Life concepts and popular computer simulation systems, namely 3D graphics engines and Newtonian physics simulation engines, with possible applications to complex systems research and digital entertainment. The theoretical model of a dual layered heterogeneous network of components is proposed as a solution for reasoning in environments where agents' perceptions are provided as arbitrarily sized vectors of symbols. A set of network components aimed at spatial reasoning is presented. Algorithms designed to maintain open evolution under real-time computational constraints are proposed. Experimental results are presented and analyzed.

Keywords

artificial life, complex systems, evolutionary computation

Subject

Artificial Life

Conference

The First IEEE Symposium on Artificial Life, April 2007