Decentralized Coordination Processes

This project reveals trends in algorithm strategy which helps to optimize the performance of independent agents in coordinated decision making problems. The software test bed constructed for this purpose models a problem consisting of independent agents which must coordinate, in the absence of communication, to achieve a common goal. The only decision making tool provided, in this scenario, is an algorithm designed to help the agents make independent decisions which can ultimately be useful in achieving the goal of the group. This project was motivated by desire to investigate the potential application of coordinated decision making algorithms to robotic surveillance missions on Mars. Thus, the problem formulation is comprised of unmanned aerial vehicles (UAVs) that have a common goal of target acquisition. We designed, implemented and tested several algorithms for success at achieving this goal. Results of this testing reveal that the impact of centralized control on algorithm performance is negligible compared to the benefits of added complexity and intelligence in coordination algorithms. This conclusion is promising for the application of coordinated decision making algorithms to robotic surveillance missions