Using Multi-Agent System Technologies in Risk Bots

We present a multi-agent architecture for playing the game of Risk which is a multi-player board game in which the players control armies and try to conquer the world through attacking each others territories. Our solution puts an agent in every territory, and let them negotiate about what actions to prioritize in the phases of placing armies, attack, and fortify. The results of a tournament of 13 participating bots, shows outstanding results for our solution, MARS which ends first or second (out of six participants) in 507 out of 792 matches.