THE FUTURE OF AIR TRAFFIC CONTROL: HUMAN OPERATORS AND AUTOMATION
TESTIMONY BY CHRISTOPHER WICKENS
The nation's air traffic control system is responsible for managing a complex mixture of air traffic from commercial, general, corporate, and military aviation. Despite the strong safety record achieved over the last several decades, the system does suffer occasional serious disruption, often the result of outdated and failed equipment. When equipment failures occur, the safety of passengers and airplanes depends entirely on the skills of controllers and pilots.
Pressures to increase the number of flights that can be moved through the national airspace system safely and efficiently have led to proposals to provide more reliable and powerful equipment and at the same time increase the level of automation in air traffic control facilities--that is, to use advances in technology to take over tasks that are currently performed by humans. Such proposals have raised concern that automation may compromise the safety of the system by marginalizing the human controller's ability to provide the necessary backup when disruptions occur.
A second concern revolves around current planning toward a concept in which pilots, airline dispatchers, and managers assume more authority for air traffic control. This concept, referred to as free flight, has many implications for the controller's performance that parallel the implications of high levels of automation.
The Panel on Human Factors in Air Traffic Control Automation was convened at the request of the Federal Aviation Administration (FAA) to study the air traffic control system, the national airspace system, and future automation alternatives from a human factors perspective. The central premise of the analysis is that considerations of public safety require that the air traffic control system continue to be designed so that the human controller can intervene successfully as spot failures in the software or environmental disturbances require or can even assume manual control when the automation fails. The panel's first phase, which focused on the current system and its development, led to recommendations regarding safety and efficiency, system management, personnel selection and training, the development of an integrated approach to human factors, and system design considerations of human strengths and vulnerabilities, which were analyzed in some detail. These recommendations and their supporting analyses are presented in Flight to the Future: Human Factors in Air Traffic Control, the panel's Phase I report. The panel's second phase assessed future automation alternatives and the role of the human operator in ensuring safety and efficiency in the air traffic control system. Along with this assessment, the panel has included a human factors analysis of free flight focusing on its implications for the performance of air traffic controllers.
The panel concludes that current system needs and the availability of various technologies provide adequate justification to continue the development and implementation of some forms of air traffic control automation, but we strongly argue that this continuation should be driven by the philosophy of human-centered automation, which we characterize as follows:
• The choice of what to automate should be guided by the need to compensate for human vulnerabilities and exploit human strengths.
• The development of the automated tools should proceed with the active involvement of both users and trained human factors practitioners.
• The evaluation of such tools should be carried out with human-in-the-loop simulation and careful experimental design.
• The introduction of these tools into the workplace should proceed gradually, with adequate attention to user requirements, to training, and to the differences among facilities.
• The operational experience from initial introduction should be very carefully monitored, with mechanisms in place to respond rapidly to the lessons learned from the experiences.
We used this definition as a backdrop and coupled it with our understanding of (1) the air safety goals for the next decade set forth in the Gore commission report, (2) the new technology available and on the horizon, and (3) research results from other related domains in which computer based automation has been applied (e.g., the flight deck, the industrial workplace). As a result, we formulated a number of conclusions and recommendations, most of which are briefly described below. The full text of these, as well as the background justification, can be found in the text of our second report: The Future of Air Traffic Control: Human Operators and Automation.
Our first recommendation addresses the question: How far should automation go? We recommend high levels of automation for functions that collect, integrate, filter and display information to the controller. Such automation, if carried out with analysis of controller information needs and an understanding of the perceptual and cognitive characteristics of the controller, can greatly improve awareness of the airspace and support effective decision making.
But for several reasons we recommend that computers should not make decisions for the controller. Instead computers should provide assistance in making those decisions, for example by offering suggestions and providing interactive planning tools. The basis of this recommendation is founded in research which reveals that when humans monitor computer decision making, rather than making decisions themselves, they are less able to detect situations in which there is a failure in the decision process, or to intervene should it be necessary for them to take over full responsibility for decision making. Furthermore there is a possibility that the controllers'decision making skills will decline, if they do not practice these skills (i.e., while the automation is carrying out the decisions).
Our second recommendation concerns emergency recovery, since emergencies do inevitably happen. The panel was concerned about a future scenario in which the use of automated tools creates a densely packed airspace, and an unexpected emergency develops. This situation requires a much greater level of direct controller involvement to recover, when time is critical because the separation between aircraft is much smaller than in today's airspace. Yet for reasons described in the previous paragraph, the controller may be less effective in intervention, if decision making has been highly automated. Research is necessary to understand much more about the characteristics and time delays of human response in such situations, and to insure that the air traffic management system is designed so that the controller skills necessary for emergency recovery are not compromised.
Our third recommendation concerns the concept of free flight, which is an alternative vision of the national airspace operation relative to the ground-based automation that was the focus of our panel. We assessed the ability of current computer simulation models to predict the future safety implications of both free flight and a ground-based automation scenario; in the latter case most authority unambiguously resides in the air traffic control facilities. For a variety of reasons, we concluded that predictions of the safety enhancement, as prescribed by the Gore Commission report, are easier to make with a ground-based automation scenario than with a free-flight scenario. This is the result of several factors that make modeling safety parameters more uncertain with a free flight scenario, in which responsibility for flight path control is shifted to pilots: (1) There are no good models to predict how two pilots might negotiate a difficult conflict trajectory, nor how long that negotiation might take in a "worst case" situation. (2) Since all free flight concepts accept that air traffic control will have authority in some circumstances (e.g., in final clearances, or in resolving conflicts that cannot be resolved by pilots), any free flight concept invites some ambiguity of authority at certain times or under certain circumstances. Such ambiguity is of great concern for flight safety. (3) Free maneuvering will create a more complex, less structured airspace, which could make it considerably more difficult for controllers to maintain situation awareness, and recover effectively in case of an emergency.
For these reasons, we concluded that extensive human-in-the-loop simulation research is necessary before confidence can be gained as to the safety enhancement potential of free flight. Therefore we recommend that until the safety of free flight is confidently demonstrated, authority should reside on the ground, and the FAA should continue to pursue ground-based automation initiatives. The panel was sensitive to the goals of increased efficiency that he behind the concept of free flight, and formulated a vision of how such goals could be achieved while maintaining most authority on the ground for non-Oceanic flight. This vision would assume the establishment of more flexible direct routing, the capabilities of more precise aircraft tracking through satellite navigation systems (replacing current radars) and extensive utilization of human centered automation tools such as CTAS and conflict probes. It would not deny pilot requested alterations of flight paths on the basis of weather, and at higher altitudes. But many such alterations would become unnecessary with the greater efficiency availed by the anticipated changes.
Our fourth recommendation concerns how the FAA should insure human factors input to all stages of automation product development, from defining its functionality, to designing its software and human-computer interface, to product evaluation and testing, to controller training, to monitoring operational field experience. We believe that this input is necessary at all stages to insure the application of human centered automation to the next generation of ATC systems, in order to meet the constraints of capacity demand and safety enhancement. The panel reviewed an internal report by the FAA regarding the Status and Organization of Human Factors within the FAA (August 5, 1996) and endorsed the conclusions of that report. Thus we recommend that a centralized human factors office within the FAA should be given the responsibility, authority, and resources to monitor the quality of human factors involvement in research, development and acquisition of all ATC products and systems, whether this is carried out within the FAA or at contractor sites (including contracted human factors specialists supporting product teams).
The application of commercial off the shelf (COTS) technology should receive as much human factors oversight as the development of new products. The panel also recommends that the FAA carefully consider the harmonization of various automated subsystems as they are introduced into the ground and air facilities, and that laboratories at NASA, the FAA Technical Center, and the FAA Civil Aeromedical Institute--all of which are repositories of strong human factors expertise--should remain closely involved with system development and fielding.
The panel highlighted the importance of some formal human factors training for all personnel assigned to product teams. The panel also recommended that operators chosen to work with new systems or subsystems should be given specific training to achieve an understanding of the principles of system operation, including the logic and algorithms underlying the system, as well as the practice o system operation.
The panel also examined the human factors issues associated with specific air traffic control automation functions related to surveillance and communications, flight information, immediate conflict avoidance, long range flight planning, and support functions (Airway Facilities and training). Some of our more specific conclusions and recommendations with regard to these issues are as follows.
Communications. The panel considered data link to be a critical element in ATC automation, that addresses one of the most important vulnerabilities in the current ATC system--errors in voice communication. It is a system about which good human factors research and analysis have been carried out. We recommend that a data link system preserve redundant channels of voice information and support graphics display where relevant.
We also recommend careful analysis and study of how data link may change the roles of both air traffic controllers and pilots in team operations, and we recommend that pilots always play an active role in monitoring uplinked data before it is entered into a flight management system on board the aircraft.
Flight data. The panel recognizes the value of electronic flight data and recommends that if the ATC system evolves toward one without paper flight strips, the functional value of those strips in current use should also be preserved in a "stripless" environment.
Immediate conflict avoidance. The panel examined the TCAS system in depth and drew from it several human factors lessons regarding the unexpected consequences of the introduction of this technology into the airspace; these lessons could be applied to other systems. We recommend that efforts continue toward making available TCAS resolution advisory information on ATC terminals. We also recommend that careful human factors evaluation be carried out on the precision runway monitor system for parallel approaches, and that possible interactions between this information provided to controllers and the cockpit information provided by advanced TCAS systems be very carefully studied. For ground-based traffic monitoring systems, we advocate that careful consideration be given to the interaction between different systems (especially those that utilize different sources of sensor information), to ensure that conflicting information is never provided to pilots and controllers regarding the status and safety of runway and taxiway paths.
Strategic long range planning. The panel endorsed the human centered automation concepts embodied in the controller TRACON automation system (CTAS), and the user request evaluation tool (URET) for conflict probe. We were particularly encouraged by the careful human factors input provided by NASA at all stages of development and field testing of CTAS at
operational facilities of Dallas-Fort Worth Airport and Denver Airport. The introduction of the URET system is less mature, and we encourage careful human factors monitoring and supervision of that introduction. Possible concerns that its use could create a more complex airspace, and a possible shift in roles between controllers, should be evaluated through such monitoring.
Support functions. We recommend that considerable attention and resources be provided to assure the application of human centered automation concepts to the development of new Airway Facilities and test equipment, and to understand the Airway Facilities specialist's changing role and skill requirements as maintenance functions are increasingly automated.
Finally, in the first volume prepared by the panel, we provided a number of recommendations pertaining to human factors issues whose impacts extend beyond automation, in areas such as the importance of team and communications training, shiftwork, workload, fielding of new equipment, and the role of simulation for training air traffic control specialists.