abstract
We propose BlastPursuit for the Multi-Agent Systems project. This innovative project aims to create an augmented version of the classic Bomberman game, showcasing advanced concepts of multi-agent systems, including agent autonomy, strategic interaction, and adaptive learning within a complex environment. Our primary objective is to develop a multi-agent system where two squads of agents, Bombermen and Killers, each with distinct goals, navigate and interact within a grid-based environment. By employing various Reinforcement Learning (RL) algorithms, we aim to investigate their efficacy in enabling these agents to develop successful strategies and autonomously make decisions in different scenarios.
