Satisficing Game Theory for Conflict Resolution and Traffic Optimization

Abstract

This paper addresses the problem of conflict resolution in an Airborne Self-Separation airspace with an approach inspired by the emerging framework of Collective Intelligence (COIN). This is a framework designed to address situations where there is no centralized control and where a clear objective function needs to be optimized. The “intelligent” behavior of the system is an emergent property of a system where each agent maximizes the optimality function for the system while interacting locally with other agents and pursuing its own goals. The paper conducts a review of some conflict resolution techniques, including a centralized approach and two decentralized strategies. A particularly promising approach is Satisficing Game Theory (SGT), which is based on the idea that by exchanging information about each agent, cooperation among agents can be realized, even in a completely distributed system. The authors develop software capable of rendering three-dimensional animations of flight scenarios using pre-defined set-ups to achieve precise rendering and to increase the number of playback options. In test scenarios, they first look at two-dimensional cases and test the robustness of the model algorithms on extremely “pessimistic” scenarios. Among the issues remaining to be tested are the overall safety performance using the SGT approach. Its robustness also needs to be studied in light of equipment failures or malicious behavior by aircraft or incomplete communications. Human factor and institutional issues must also be addressed in greater detail.