Machine learning-based go-around predictions have been discussed in the research community for some years. Much work has been done developing algorithms and testing their accuracy, motivated by the assumption that time-in-advance information on the go-around likelihood of arrival aircraft will benefit air traffic controllers. The question of how to incorporate predictive and probabilistic information into the operation and how to evaluate their operational impact has yet to be investigated. This paper presents a first step toward assessing the operational impact of a machine learning-based decision support tool. Therefore, a low-fidelity, human-in-the-loop simulation exercise with air traffic controllers discovers potential new tactics enabled by a go-around prediction tool and evaluates them regarding safety, resilience, and capacity.     «

Machine learning-based go-around predictions have been discussed in the research community for some years. Much work has been done developing algorithms and testing their accuracy, motivated by the assumption that time-in-advance information on the go-around likelihood of arrival aircraft will benefit air traffic controllers. The question of how to incorporate predictive and probabilistic information into the operation and how to evaluate their operational impact has yet to be investigated. This...     »