Abstract
Efficient collaboration within a Multi-Robot System (MRS) is crucial for managing complex tasks in ever-changing environments, considering various physical constraints. To investigate potential collaborative strategies for MRSs, this research utilizes the RoboCup Rescue Simulation (RCRS) testbed that supports programmable rescue agents in disaster mitigation settings to explore and refine AI-driven team strategies. We delve into a teamwork-centric method tailored for the RCRS challenge, integrating a task modeling framework into the simulation to achieve a better breakdown of tasks and to manage the collaborative requests among the platoon agents. Additionally, a communication protocol was developed for the ability of agents to adeptly share and utilize information dynamically in fluctuating environments. This research study demonstrated an improved overall simulation performance compared with the default agents in complex map environments. The approaches developed in this research are promising to bridge the effective MAS collaboration mechanisms with the practical constraints of physical multi-robot systems, potentially enhancing cooperative endeavors in real-world scenarios.