Flexible multi-agent algorithm for distributed decision making

This article presents an algorithm for decentralized control and decision making among a number of unmanned aircraft systems (UASs). The approach is scalable and adaptable to a variety of specific mission tasks. Additionally, the algorithm could easily be adapted for use on land or sea-based systems. Each UAS involved in an automated decision making process uses a selection of both predefined and derived parameters with corresponding predetermined weights to develop a score. These scores, and associated “scorecards,” are then iteratively distributed to all involved agents, ranked, and used to determine task participants. Simulation as well as flight testing shows the algorithm provides appropriate team assignments in a variety of situations while remaining robust to agent dropouts, inconsistent communications, and dynamic situations.