This paper describes the implementation of a soccer team for the RoboCup Simulator League using techniques based on Swarm Intelligence. The main idea of using Swarm Intelligence is to simplify the codification of each agent (player) and, at the same time, to get complex cooperative behaviors using very simple heuristics. In addition to describe the general structure of each agent, this paper also analyzes the main algorithms implemented by our swarm strategies. Although the size of our code is very small, the evaluation of our soccer team demonstrates that robust and competitive performances can be expected from the proposed
methodology. Finally, this paper describes our plans to extend the functionality of the current team in order to make it as good as previous champion teams.
|