Interactive Evolution of Camouflage

Computational models of complex natural systems, even if quite abstract, can help to understand the natural system and can lead to new bio-inspired computer techniques. This talk describes an abstract model simulating the evolution of camouflage due to vision-based predation. For a given background image, it synthesizes camouflage patterns which can be difficult to find on the background. Inspired by an aspect of evolutionary biology, this model combines procedural texture synthesis with computational techniques for optimization and self-organization. This initial stage of the model is a hybrid system making use of "human computation" for the predator's vision.

Speaker: Craig Reynolds is a Senior Researcher at Sony Computer Entertainment's US R&D group in Foster City, California (www.research.scea.com). His recent work involves evolutionary texture synthesis. Earlier projects focused on autonomous characters, including models of collective construction, PSCrowd (an high performance crowd simulator for PS3), OpenSteer (an open source library of steering behaviors) and "boids" (a model of flocks, herds and school). He has previously worked on animation and game production, plus developing tools for both fields, at: DreamWorks, Silicon Studio, Electronic Arts, Symbolics and Information International Inc. He won a Scientific and Engineering Academy Award in 1998 for "pioneering contributions to the development of three dimensional computer animation for motion picture production." (http://www.red3d.com/cwr/)