Evaluation of Levels of Automation for Non-Normal Event Resolution 2011-01-2673

Increasing the automation in a flight deck has not always led to benefits to the pilot; instead it is often found that inappropriate implementation of automation can cause a variety of issues including loss of situation awareness, poorer performance, and a perceived increase in workload. It is important to determine in what context a pilot might benefit the most from the inclusion of automation and then to determine the appropriate level at which that automation is implemented.

The objective of this study was to evaluate what level of automation is most appropriate for supporting the pilots in the performance of non-normal event resolution tasks. Pilots were given a series of scenarios where the level of automation for performing tasks was varied from no automation to full automation. A related objective was to determine whether the use of a voice interface is effective in providing sufficient feedback to support the pilots' situation awareness. In the scenarios, the number of computer-generated voice messages was varied to provide more or less information about the automation's performance.

Subjective workload ratings indicated a solid trend for lower workload ratings for those conditions where the pilot had significant support from automation (i.e. when automation performed the tasks). When given the opportunity to design the automation for themselves, almost all pilots specified that the automation should perform the tasks. The pilots also requested to keep the voice messages for some, if not all, tasks and were evenly split over whether the automation should request permission or not before performing the tasks. The results of this research suggest that in order for increased automation to be an effective aid to the pilot, it needs to be implemented in such a way that the pilot has insight into, and some degree of control over, the system's actions.