Open-Ended Evolution in Visually Rich Virtual Worlds: Implementation, Analysis, and Use in Undergraduate Education

Hampshire College, Amherst MA

Investigators

Abstract

Evolutionary computation systems, in which Darwinian principles of variation and selection govern the development of digital representations in simulated environments, have applications in several areas of science and engineering. With the addition of facilities for high-quality 3D visualization and interactivity, such systems can also provide dramatic windows into complex and dynamically evolving virtual worlds. This project builds on the PI's work on evolutionary computation to produce software that implements open-ended evolution in visually rich virtual worlds. The project aims to produce software that provides a laboratory environment of research opportunities for fields ranging from evolutionary biology to optimization theory. This software simulates evolutionary processes and evolves based on Darwinian principles, thus improving itself. First goal of the project is to attract and engage undergraduates (both science majors and non-majors) via the intrinsically compelling life-like dynamics of behaviorally complex, visually presented virtual worlds. To achieve this, the developed software includes a graphical user interface that enables non-programmers to formulate and investigate meaningful questions about evolutionary processes, providing genuine research experience to undergraduates. Second goal of the project is to support fundamental research on these same questions. The system uses modern, robust evolutionary computing technologies and can be implemented both for single PCs and for high-performance computer clusters. The software supports both undergraduate and faculty research, and contributes to the understanding of evolution, by developing useful technologies based on evolutionary computation, and by presenting research results in formats understandable by non-scientists.