Niching agents improve control success in complex video game environments
Niching Agents in The Core
Artificial Intelligence
Summary
Evolving smart agents to control characters in video games is hard without clear goals to measure how good they are. The Core is a way for agents to get better by competing against each other, but it doesn’t use traditional scoring methods. This paper shows that if agents are specialized or 'niched' to smaller parts of the game environment, they perform better than agents trained to handle the whole game or different parts. This helps create smarter game controllers without needing a strict fitness score.
What this means in practice
- •For game ai developers: Improve game character control systems by evolving agents specialized to parts of game environments, enhancing performance over generalist approaches.
- •For robotic control engineers: Use niching strategies in evolutionary control to create more efficient autonomous system controllers tuned to specific operational scenarios.
Authors
Gary B. Parker, Jim O'Connor, John Asaro
Abstract
The Core is a unique competitive co-evolution algorithm that allows agents to evolve autonomous control without utilizing a traditional fitness function. The agents evolve via local interactions through tournament selection, crossover, and mutation, producing offspring by evolving better controllers. Previous works have shown The Core's ability to evolve agents capable of combat and navigation in the Xpilot video game. This research expands upon that premise by niching agents to specific subsets of the original environment The Core was tested in. Our results demonstrate the niched agents capacity for success over agents niched to the entire system and agents niched to different sub-environments.