Christopher D. Wickens University of Illinois at Urbana-Champaign
The North American air traffic control system has been an incredibly safe one, coordinating roughly 60 million operations (takeoffs and landings) per year; and not having been directly associated with a single fatality resulting from mid-air collisions, during that last 40 years. Such a record of safety is attributable to the skills of air traffic controllers, and to the inherent redundancies that are built into the system. At the same time, increasing demands for air travel are projected to place excessive demands on the capacity of the ATC system over the next 20 years, even as the current system appears to be incapable of meeting such demands ( Aviation Week & Space Technology, 1998). In large part, such limitations are related to antiquated equipment, ( Perry, 1997) as well as to the resolution of current radar systems limiting their ability to precisely identify the 3D location of aircraft. Three general approaches have been proposed to address these needs: modernization by which older, less reliable equipment will be replaced by newer equipment ( Aviation Week & Space Technology, 1998); automation, by which computers will assume responsibility for certain tasks, previously performed by controllers ( Wickens et al., 1998); and free flight, by which greater authority for maneuvering and traffic separation will be provided to pilots ( RTCA, 1995a,b; Planzer & Jenny, 1995).
In late 1994, a panel was convened by the National Research Council, at the request of the House Aviation Subcommittee and the FAA, to examine the second of these issue--air traffic control automation--, and in the third year, the panel was also asked to examine the issue of free flight, from the controller's perspective ( Wickens, Mavor, & McGee, 1997; Wickens, Mavor, Parasuraman, & McGee; 1998). The current paper highlights the conclusions of the panel regarding three critical human factors issues in the automation of the air traffic control system. While I provide here my own interpretation of these issues, I also acknowledge the major contributions of the co-editors of the panel reports, Anne Mavor, Raja Parasuraman, and Jim McGee in particular, and the remaining panel members in general1. I believe that my words below are quite consistent with the overall views of the panel.
All three of the major conclusions discussed below are anchored in roughly three decades of human performance research in automation, the focus of the present conference, and these conclusions clearly point to the relevance of this research to some of the most important safety issues confronting our society -- the safety of air travel -- even as they also lay out an agenda for more research, particularly addressing the concept of trust, and the human response to infrequent events.
Our panel spent three years studying air traffic control systems and learning about their details, benefiting from the expertise of two experienced controllers on the panel. We performed a cognitive task analysis, considering the delicate balancing acts that controllers must carry out between safety and efficiency; between ground and air concerns; and between following procedures to deal with routine situations, and using some creative problem solving to deal with the novel, unexpected events. We examined the cognitive vulnerabilities that should be supported by automation, as well as the cognitive strengths of controllers, that should not be neutralized or replaced by automation ( Wickens, Mavor, & McGee, 1997).____________________