Game Theory Based Decentralized Multi-Task Allocation Algorithm for Multi-Agent Systems

This paper addresses a task assignment problem to organize a flock of autonomous vehicles (AUVs) by proposing a game theory-based decentralized algorithm. The algorithm employs a market-based mechanism to assign tasks with the help of a k-means clustering method for task selection. In contrast to the conventional algorithms widely used in multi-agent systems (MAS), the shortest path among a set of tasks has been obtained within the cluster using the ant colony optimization (ACO) algorithm. Additionally, unmanned aerial vehicles (UAVs) have been adopted in this study due to their effective manoeuvrability and field of view (FOV). To test the algorithm's effectiveness and facilitate the equations, the 6-DOF quadrotor model was developed in MATLAB/SIMULINK. Several scenarios have been carried out for various environments. As per the results gathered from the simulations, it is demonstrated that the proposed algorithm produces excellent solutions in small and medium-scale applications, and in large-scale applications, it indicates provable worst-case performance.