COMPARING CENTRALIZED AND DECENTRALIZED MULTI-AGENT APPROACHES TO AIR TRAFFIC CONTROL

摘要

Air traffic has been increasing in the last decades, representing a major mode of transportation. The FAA (Federal Aviation Administration) and the aeronautical industries are both predicting a growth between 150% to 250% over the next two decades. In this paper, we present and compare centralized and a decentralized agent-based approaches to autonomously manage air traffic. Agents assuming the roles of aircraft and that of the air traffic controller detecting possible mid-air collisions and proposing new routes for the aircraft to avoid such collisions, while attempting to minimize the cost of the deviations. During simulations we were able to avoid collisions between aircraft from a minimum maneuver area of 12.25 km2 per aircraft and identified that the centralized approach has a better performance, when compared to the decentralized approach.