Carl Westin

4 Introduction the mental workload associated with controlling and separating traffic. 1, 2 The intro- duction of more advanced automation is considered necessary to overcome current traffic delays and achieve future ATC capacity goals. For example, in SESAR, the European air traffic management (ATM) community is working towards achieving a three-fold increase in airspace capacity between 2005 and 2020, without adding more controllers. 3 Similar goals have been established by other initiatives world- wide, such as NextGen in the United States. 4 These targets require the use of more sophisticated automation that supports and eases the cognitive burden of the controller in problem solving and decision-making tasks. 5 As a critical part of this, automation is foreseen to assume a greater tacti- cal role in the short- and medium-term timeframe of planning and executive ATC. Moreover, automation is likely to act more as an adviser providing solutions to the controller in regards to, for instance, airspace reconfiguration and planning traffic. Such automated decision aids are expected to be especially beneficial in the strate- gic and tactical phase of separating traffic by alleviating controller workload. 6, 7 Currently, controllers carry out this key task of conflict detection and resolution (CD&R) largely manually and with limited decision support. 1-2 Problem definition Of pivotal importance for the future ATC system is that the automation developed is accepted and used by the controllers it intends to benefit and support. Unfortunately, in the past decades several CD&R decision aids have been rejected or used in ways not intended by the designer. 8–10 Rejection of automation has often been attributed to large uncertainties in CD&R algorithms, leading to inaccuracies in conflict de- tection and unreliable resolution advisories. 10, 11 Automation disuse has also been attributed to inappropriate decision thresholds that lead to either an extensive num- ber of false warnings or failure to detect conflicts. 12, 13 The acceptance, or reliance, on automation is believed to be affected by various factors influencing operator at- titudes toward automation, such as trust, perceived risk, perceived reliability, level of automation, age, and job satisfaction. 13–16 Possibly, the observed acceptance issues can be attributed to differences in con- flict resolution strategies between the automated decision aid and the controller. Since there are often several alternative solutions to a conflict, the automation and human do not necessarily agree on which one to apply. For example, in re- cent human-in-the-loop simulations exploring decision aiding automation for con- trollers, the automation was perceived as occasionally ‘fighting’ against the con- troller on how to solve conflicts. 17 In a study investigating the adoption process of a conflict detection system called URET (User Request Evaluation Tool), Bolic