Adaptive Automation to Improve Human Performance in Supervision of Multiple Uninhabited Aerial Vehicles: Individual Markers of Performance

Abstract

Adaptive automation has been shown to offer flexible, context-dependent, and user-specific automation that can enhance human-system performance. While several invocation methods for adaptive automation have been proposed and tested in experimental settings, it is not clear which of these methods can practically be implemented in operational environments. It is therefore important to explore measures that are both predictive of individual performance and that can be easily administered in actual work environments. This study examined the efficacy of using both baseline manual performance and working memory capacity to predict future performance with automation. Participants were assisted by context-dependent adaptive automation during a simulated command and control task. Results showed that baseline performance without automation predicted overall human-automation performance. Working memory capacity did not predict overall performance, but did predict effective use of the automated aids, so that participants with higher working memory scores used the aids more effectively. These results suggest that effectiveness of human-automation teams can be predicted with quick, cost-efficient, easily measureable markers of performance and can therefore provide practical invocation strategies for adaptive automation.