Monday, March 1, 2010
This game was developed to research artificial intelligence and it shows evolution in action. The game is NERO, which stands for Neuro-Evolving Robotic Operatives. In the game, the characters evolve to improve their performance of tasks.
It’s based on the neural network research in the laboratory of Risto Miikkulainen, a professor in the Department of Computer Science and a member of the BEACON team.
In NERO, players command armies of robot-soldiers and train novice soldiers in the skills needed for battle.
This is how I described it in a story about the game:
The soldiers can be trained to go after the enemy, avoid the enemy, run around walls, navigate through mazes and perform other tasks. The soldiers’ learning continues to grow during the game. The best ones reproduce more well-trained soldiers while those that perform poorly are eventually replaced.
BEACON (it’s an acronym for Bio/computational Evolution in Action CONsortium), is a $25 million, multi-university center established by the National Science Foundation (NSF) that will study evolution in action in natural and virtual settings.
It’s a joint effort by Michigan State University, The University of Texas at Austin, the University of Washington, the University of Idaho and North Carolina A&T State University.
Rather than study evolution by looking at the past (fossil records or comparing DNA among species to discover common ancestry), BEACON scientists will focus on evolution as an ongoing process.
They will use organisms in laboratories and at field sites, and use “digital organisms” undergoing real evolution on computers.
“The goal of the project is to transfer knowledge about biological mechanisms to engineering and computer science,” Miikkulainen said. “On the other hand, we will use digital evolution to gain insight into biological phenomena that would be difficult or impossible to study in the laboratory.”I asked the professor if BEACON would have anything to do with NERO-related research.
“Absolutely,” he said. “We are actively working on OpenNERO (an expanded, open source version of NERO, intended as a general research and education platform for artificial intelligence). We plan to use OpenNERO as a platform for several research projects in BEACON, and other groups under BEACON probably will as well.”
Miikkulainen said Michigan State researchers have a platform called Avida that’s based on evolutionary programming of simple digital organisms.
“In NERO,” he said, “the agents are much more complex (mobile, interacting teams), and they are controlled by neural networks. So where Avida may be used to study low-level issues, NERO is better suited for studying high-level behavior. So they should complement each other nicely, and also perhaps some of the features can be integrated as well.”
Other University of Texas at Austin researchers working on the BEACON team are: David Hillis, Lauren Ancel Meyers and Claus Wilke (Section of Integrative Biology) and Andrew Ellington (Department of Chemistry and Biochemistry).