Multi Agent Co-operative Learning (MACL) is a project that looks at ways in which agents learning to perform tasks in a cooperative manner. This includes using methods such as Reinforcement Learning.
Galcon is a simple game where the objective is to capture enemy and neutral planets by destroying enemy fleets. The winner is a person capturing the most number of planets or completely destroys all enemy planets. Planet Wars is an implementation of this game by Google, provided for an AI Challenge. The framework provides a way to program bots and test them against a set of standard bots available. We use it to test our methods and suggest is as a general framework to solve multi-agent problems.
To get more information on planet wars, checkout their page http://planetwars.aichallenge.org/problem_description.php