Evaluation of a Model of Expert Decision Making in Air Traffic Control

Abstract

The resolution of conflicts between aircraft is regarded as a very complex optimisation problem, yet air traffic controllers are able to perform the task with high rates of success under demanding circumstances. Much of the expertise of the air traffic controller appears to lie in the ability to select the appropriate strategy for the problem. In this project, we describe a model of expert decision making for the air traffic control conflict resolution task. A key assumption of the approach is that controllers adopt relatively simple heuristics to solve the complex trajectory problem. We model expert decision making as a serial search process in a hierarchical tree, in which the selection of a decision option for further evaluation is constrained by the situation. This is part of a broader project that is developing a new approach for simulating the tasks that a human operator performs, and the workload that the human experiences while carrying out those tasks. In this paper, we will present an analysis of the behaviour of an initial version of this model. The key aim of this analysis is to compare the model's behaviour against the behaviour of expert controllers under varying scenario complexity. The analysis relies on both data from our model simulation runs and recordings of the activities of 14 air traffic controllers. It is based on static aircraft scenarios of varying complexity.