abstract
The aim of this project is to simulate the Cops and Thieves game through a multi-agent system. The Cops and Thieves game involves players assuming the roles of either cops or thieves, where thieves aim to collect coins while evading capture by cops. We will employ Jason and the Q-Learning reinforcement learning algorithm to model the game’s dynamics effectively. Jason, as a platform for multi-agent system development, provides the essential infrastructure for creating intelligent agents capable of strategic decision-making and interaction. Furthermore, the Q-Learning algorithm, renowned for its ability to facilitate agents’ learning from experience and optimizing their actions, aligns seamlessly with the project’s goal of enabling intelligent decisionmaking among the game entities.
The game dynamics will involve randomly generating teams of cops and thieves, positioning safe zones and coins, and determining the playing area size based on the number of players. Thieves will navigate the playing area, including safe zones, to collect coins, while cops will patrol the area to apprehend thieves. The game concludes when either all coins are collected by the thieves, resulting in their victory, or all thieves are apprehended by the cops, leading to the cops’ triumph.
outcomes
