Human performance and workload in future air traffic management: Decision support tools for controllers

The current National Airspace System (NAS) is facing pressures to handle a steady growth in air traffic that is projected for the future. Psychologists' are needed to provide the cognitive background for how air traffic controllers (ATCos) process information and how system changes could affect the process. The present work was designed to investigate the human performance consequences of the transition into a shared decision-making system. This is of theoretical interest because models of human-automation interaction postulate the need for feedback to the operator on automation states (Parasuraman et al. 2000). The practical relevance is driven by the need for developing effective and safe future air traffic management (ATM) systems which can transition smoothly from the current system.; Metzger et al., (2003) and Corker, Fleming, and Lane (1999) found evidence of negative consequences of ATCo performance when ATCos handled freely maneuvering aircraft. In the present work, the effects of traffic predictability, and reliable and imperfect automation on ATCo performance, attention allocation, and mental workload were investigated. It was hypothesized that controller performance would improve, mental workload would decrease, and ATCos would allocate attention differently to different equipped aircraft. Additionally, it was expected that ATCos would show better performance when they were supported by automation. It was expected that when the automation was imperfect, operator conflict detection performance would degrade. It was hypothesized that miss prone versus false alarm prone automation would affect eye movements and behavior differentially.; Results from two experiments on the role of automation and attention allocation in human performance showed that: (a) ATCos detected conflicts earlier when freely maneuvering aircraft provided the ATCo with intentions regarding their proposed new trajectories; (b) reliable automation improved performance compared to manual control and imperfect automation degraded performance supporting previous work with automation in an ATM environment (Metzger & Parasuraman, 2005); (c) optimal distribution of attention was better with miss versus false alarm prone automation. The knowledge gained may be applied to future ATM concepts as well as the design of support tools that complement the operator while taking into account the costs related with types of automation imperfections.