BEACON Researchers at Work: Evolution with video games

This week’s BEACON Researchers at Work blog post is by MSU graduate student Leigh Sheneman.

LeighMost all of us remember sitting around thinking up wild possibilities of what our future would hold. For me, this daydreaming took place in the early 80s in the middle of nowhere Texas. Since there was little to do, I spent a lot of time playing computer games. Thus, my top “this could never happen, but it would be cool…” statements of my childhood were: a) to have my name associated with a computer game and b) to always be able to ask why without getting in trouble. Now these both seem simple (at the time these were BIG, I promise), so simple that I actually do them on a daily basis.

How, you ask, do I get to program a video game while doing research? The answer is EvoSphere, a digital evolution package that is built using the same technology used to create your favorite video game. The scientific goal of the project is to co-evolve the bodies and brains of digital organisms in a 3D digital world, and to explore what is similar or different among different evolving systems in a common environment.

However, since EvoSphere is developed using a gaming engine, people outside the evolution and/or programing fields can boot up their favorite gaming device and watch evolution in action! Our youth are engrossed in the world of video games. With EvoSphere we are able to bridge a world that people already know with scientific ideas to inspire more interest in STEM research.

While all this exciting outreach is taking place, the researcher in me can use the very same system to dig into the all-important question of “Why?” On that note, EvoSphere’s code base is extremely modular, meaning you can easily develop a new environment for your test organisms to explore. Further, the organisms are modular too. As researchers this gives us the ability to give multiple brain types to the same body form, or vise versa. This is demonstrated nicely in a video we submitted for the Virtual Creatures Competition at GECCO 2014.

As the video demonstrates, we currently have three working brain types: Markov (coded by Dr. Arend Hintze), NEAT (Dr. Joel Lehman) and a genetic programming (Dr. Stephan M. Winkler). Dr. Hintze and David Phillips worked together on the three body structures available. Any brain will work with any morphology. I joined the group, led by Dr. Rob Pennock, in May and have added the sensing abilities that make the food eating experiment in the video possible.

I could expand on EvoSphere’s ability to evolve 100s of organisms at a time or the different ways we visualize changes over time, but the amazing advances in digital evolution the program brings are much more exciting. By linking the brain and morphology we are able to monitor how evolution changes both body and behavior over time. Additionally, the linkage between brain and body adds tons of complexity, for free! Which is huge, as a lack of complexity is a limiting factor in current research.

The group continues to refine the platform with more brain and body types. We hope to soon have an Avidian brain available through the efforts of Mike Wiesenauer. We are also committed to enhancing the ability to analyze the behaviors of organism in real-time. Our hope is to offer a platform that other labs can perform research in without a steep learning curve. As the community grows so will the number of brain and morphology types as well as possible environments. The plug and play nature of EvoSphere makes it easy to share experimental components with each other.

Our goals are scientific, but this work connects to my daydreams of youth. As the child who frequently got lost in video game titles such as Doom and Heretic, I also see how growing the EvoSphere community may one day allow gamers to experience a new level of play. For those who play as awful as I do, the creatures we encounter could evolve themselves to an easier level of play. The same exact opponents could evolve more difficult behaviors for those gamers who master objectives more efficiently. As you can see, to adapt enemies to the match the abilities of the player will bring the gaming industry to a whole new level. Once this circle is complete, my “this could never happen, but it would be cool…” statements will have been met beyond my wildest imagination!

For more information about Leigh’s work, you can contact her at leighs at egr dot msu dot edu.

This entry was posted in BEACON Researchers at Work and tagged , , , . Bookmark the permalink.

Comments are closed.