DEEP LEARNING APPLICATIONS FOR HUMAN-CENTERED AIR TRAFFIC MANAGEMENT

<p dir="ltr">In human-centered Air Traffic Management (ATM), Air Traffic Controllers (ATCs) monitor and command the aircraft using radar, computers, or visual references. However, the growth of air traffic volume has put a serious challenge on the safety and efficiency of the ATM system due to the increased workload for ATCs. In 2021, the International Civil Aviation Organization reports the total number of passengers carried on scheduled services has reached 2.3 billion, a 28.1 percent increase from 2020, despite the impact of COVID-19. Thus, developing automated assistant tools for ATCs is crucial, and doing so by using data-driven methods that analyze and study the historical flight data becomes more and more popular due to the advancement of data collecting and processing technologies. On the other hand, despite the increase in the level of autonomy, ATM will still be a human-centered task in the near future. Even in the urban ATM where most aerial vehicles are unmanned, FAA has indicated that unmanned aircraft system operators are responsible to almost all responsibilities, including avoiding hazards, reporting status to other entities, maintaining separation with other vehicles, etc. Thus, decision support tools will eventually be integrated into a human-centered system, but without a good system design, they can even degrade the system’s performance and safety. In this regard , this thesis focuses on the development of novel deep learning algorithms for 1) improving ATM safety and efficiency and 2) exploring human's cognitive states in ATM. The algorithms can work cohesively in an intelligent ATM system used by both ATCs and automated data service provider in the future advanced air mobility. </p>