Evolving Virtual Creatures: The Definitive Guide

Alex J. Champandard on February 22, 2008

AI research ties into games and simulations in many ways, but one of the most fascinating is the evolution of artificial life. Here’s a compilation of the best videos and white papers about applying genetic algorithms to generating the morphology and behavior of virtual embodied creatures in 3D worlds.

A big thanks goes to Tim Taylor (with whom I shared a lab during my Master’s year at the University of Edinburgh) for his page on Evolution of Morphology and Behavior for Physically Modelled Creatures, from which some of the following links are derived.

Karl Sims’ Evolved Virtual Creatures (1994)

Karl Sims research kicked off this whole area of research in A-Life, and is still a solid reference even after 14 years.

Evolving Virtual Creatures
K. Sims
Computer Graphics (Siggraph '94 Proceedings), 1994.
Download PDF
Evolving 3D Morphology and Behavior by Competition
K. Sims
Artificial Life IV Proceedings, MIT Press, 1994.
Download PDF

See the following website for more information:

A New Step for Artificial Creatures (2007)

Nicolas Lassabe’s recent work builds upon Sims’ original research.

A New Step for Artificial Creatures
N. Lassabe, H. Luga, Y. Duthen
Artificial Life, IEEE, 2007.
Download PDF

You’ll find more footage and information on these sites:

Evolving Locomotion in Embodied Agents (2004)

This research by Gene Ruebsamen uses a more biologically accurate model of neural network.

Evolving Efficient Locomotive Strategies in Embodied Agents
Gene Ruebsamen
MSc Thesis
Download PDF

Go to the following website for source code and further details:

Wing-Flapping Animats (2007)

This research from the Animat Lab tackles the problem of flight from wing-flapping.

Evolution of the Morphology and Kinematics in a Flapping-Wing UAV
de Margerie, E., Mouret, J.-B., Doncieux, S., and Meyer, J.-A.
Bioinspir. Biomim, 2007.
Download PDF

Find out more about the project itself, including more videos:

An Improved System for Creature Evolution (2006)

Thomas Miconi has a multitude of papers on this topic.

An Improved System for Artificial Creatures Evolution
T. Miconi and A. Channon
Simulation and Synthesis of Living Systems (ALIFE X), 2006.
Download PDF
In Silicon No One Can Hear You Scream: Evolving Fighting Creatures
T. Miconi
EuroGP, 2008.
Download PDF
Evolution, Complexity and Progress in Nature and in Computers
T. Miconi
Ph.D. Thesis
Download PDF

You’ll find many more videos and even source code from this page:

Hierarchically-Modular 3D Locomoting Genobots (2001)

This research combines L-Systems and genetic algorithms together for generating the skeletons and behaviors of the bots.

Evolving L-Systems To Generate Virtual Creatures
Hornby, G. S. and Pollack, J. B.
Computers and Graphics, 2001.
Download PDF

The approach is particularly well documented, and includes source code. For footage and white papers, see the following sites:

A Compilation of Evolved Creatures

MorphEngine: The Virtual Construction and Evolution Kit

Josh Bongard’s work has been used in multiple projects generating motion with evolutionary algorithms.

Incremental Approaches to the Combined Evolution of a Robot's Body and Brain
J. Bongard
Ph.D. Thesis
Download PDF

Download the toolkit from this page:

Intelligent Motion Control with an Artificial Cerebellum

Biped Tripping

Russel Smith’s work has inspired many of the papers above too, and often also provided code via the ODE physics engine.

Intelligent Motion Control with an Artificial Cerebellum
Russel Smith
Ph.D. Thesis
Download PDF

For more information, see the following websites:

Framsticks: 3D Evolution and Simulation

To wrap up, Maciej Komosinski and Szymon Ulatowski have built a great framework to allow you to build these kinds of behaviors.

What do you think about this kind of research and its applicability to games? Do you have any other links, papers or videos to share?

Discussion 1 Comments

Christoph Vlad on March 6th, 2008

It's awesome!

If you'd like to add a comment or question on this page, simply log-in to the site. You can create an account from the sign-up page if necessary... It takes less than a minute!