Trust and Human Performance in Automated Formation Flight Station-keeping

This thesis primarily describes performance and decision heuristics of human operators intervention with an autonomous formation flight (AFF) system during monitoring of a station-keeping display. Due to mental and physical workloads, automation technologies have been applied to formation flight for precise station-keeping and resultant fuel reduction, shifting control authority from humans to machines. Accordingly, the human is not directly in the control loop, but just supervises whether or not the automation works as intended. One critical problem in AFF supervisory control is that the human pilot needs to intervene with AFF system when the automated systems malfunction or their functions degrade. Thus while monitoring a station-keeping display, operators should minimize incorrect decisions for safety and cost reduction. To examine design issues in such a display, a simulation was constructed that simulated two different control systems as well as the impact of different angles of bank. 20 subjects participated in the monitoring task simulation of the station-keeping display. During the experiments, subjects were asked to intervene with AFF system when the AFF system failed to keep the trailing aircraft in the vortex area. Subjects made the most incorrect decisions when the AFF system was operated with the oscillating controller and high angle of bank. Trust of the human in the AFF system was found to be influenced by the damping ratio of the AFF controller. Most significantly, results showed that humans developed biased decision criteria to execute interventions because velocity feedback of the wing tip on this display was not adequately provided. Thesis Supervisor: Mary (Missy) L. Cummings Title: Boeing Assistant Professor of Aeronautics and Astronautics