"As long as I'm in control...": pilot preferences for and experiences with different approaches to automation management

Abstract

To date, attempts to eliminate observed problems with human automation interaction have focused primarily on improving system feedback and on modifying operator training. The effects of different forms of human machine coordination on joint system performance have received much less attention. The paper reports on the results of one of the first studies to examine pilots' preferences for and experiences with different automation management strategies and implementations. "Glass cockpit" pilots were asked to rank order and explain their attitude towards five different implementations of a future automated system in terms of their effectiveness and desirability for handling 15 different datalink/free flight scenarios. Overall, pilots expressed a strong preference for a management-by-consent approach where the automation cannot take action unless and until explicit pilot consent has been received. However, high time pressure and workload as well as low task criticality were found to shift pilots' preferences towards a management-by-exception approach where the automation is allowed to take action on its own but can be overridden by the pilot. The results of this study can serve as input to modifications of current cockpit systems and, more importantly, help designers make more informed decisions about the role and the degree of authority of automation in future system developments.