Automation and Systems Issues in Air Traffic Control

When air traffic control began about fifty years ago, its main promptings were associated with the Second World War and the consequent need to fly at night or in poor visibility (Adair, 1985). Before then, aerial navigation relied on distance and time estimations, and on the judicious selection of distinctive features and landmarks along the intended route in order to verify position and progress. Ground support, mainly in the form of radio communications, did not constitute true control over air traffic, but was a service confined to regions around airports, as accounts by pioneer aviators make clear (e.g., Saint Exupery, 1939).

One of the goals of the Institute was to help the participants to question their assumptions about air traffic control (also see Hopkin, 1991). Assumptions are critical to all decision making and problem solving because they have a hidden, yet dramatic, impact on conclusions. Assumptions determine which data are accepted or rejected as relevant to the decision process. They determine the rules that will be used to process the accepted data. They determine the relative weight each datum gets. Thus, assumptions, hidden as they may be, directly impact whether a solution is accepted or rejected.

Implementation of the FAA Advanced Automation System will be the first major human interface change in the air traffic control system since the original automation program. Replacing flight progress strips with electronic displays will require significant human engineering effort. The implementation of more complex safety-related functions also will require considerable design efforts due to the system’s inherent reliability and availability requirements. New commands and procedures, in addition, will be necessary. This report describes the FAA developmental approach of using user teams to ensure an effective computer-human interface.

Exactly fifty years after the initiation of air traffic control (ATC) in Canada, Transport Canada has launched a comprehensive and innovative program, the Canadian Automated Air Traffic System (CAATS) Project, to completely revamp the national ATC system. This paper provides a brief overview of the background, objectives, design, and implementation of that project. The intention of the paper is to present a general description of the CAATS concept, without detailing either the engineering requirements or the developmental program. Excellent reviews of some of the issues, particularly those concerning human factors engineering, that must be addressed in a comprehensive ATC automation project already have been provided by Benel, Dancey, Dehn, Gutmann, and Smith (1989) and by Simolunas and Bashinski (1991). An introduction to the concept of human error in the context of automation has been outlined by Stager (1991).

In this paper I will give you a very brief description of the UK ATC system, with special emphasis on Military Air Traffic Operations. I will describe the important landmarks in development and stress the implementation aspects and the lessons learned.

The Guild of Air Traffic Control Officers is a purely professional body with no political or trade union affiliation. It is pledged to defend and improve the technical and professional standards of the air traffic control (ATC) service within the United Kingdom. Its 1500 members include civil and military operational and non operational controllers working for various employers in the UK and other countries. Additionally, our life, associate, and corporate members include many electronics companies, airlines, private air traffic control agencies, and individuals.

In this essay, the focus is on systems issues as they relate to air-traffic control (ATC). The first issue concerns the extent to which automation should be applied in ATC. The second issue refers to the ability of the present ATC information system to provide efficient and effective control in a rapidly expanding, complex air space environment. How can our present capability in obtaining information be extended to satisfy the need for greater human cognitive competence demanded by the environment? The third issue is to determine how best we can reap advantage from our knowledge of ATC failures in designing future ATC systems.

During flight, the crew of an aircraft frequently communicates with stations on the ground. They may seek authorization for flight maneuvers, obtain information necessary to avoid collision, receive an update of weather conditions ahead, or question the operational status of navigation aids en route or at destination There is a continuous invisible link between the aircraft and the ground stations, and among the ground stations themselves. Many ground facilities and supporting services are needed for the safe and efficient operation of aircraft. To achieve harmonious functioning of all these ground facilities and services, international standardization is necessary.

The advent of unprecedented worldwide growth in air traffic, together with forecasts for continued increases at a rate which will double today’s levels by the year 2000, have jolted governments and the aviation industry — airspace users, air traffic authorities, and fare paying passengers — into the realization that today’s air traffic management mechanisms can cope neither with today’s air traffic demand nor with that of tomorrow.

Aviation activity during the coming decade is expected to grow at about the same rate as the general economy. Aviation will continue to dominate all other transportation modes in the commercial intercity passenger market. Regional/Commuter aircraft activity and the business use of general aviation is expected to experience greater growth than the larger, established airlines and personal use of general aviation (Federal Aviation Administration, 1988).

The field of environmental engineering encompasses a wide spectrum of subjects and problems. Among these issues, water and wastewater treatment and water distribution systems are the areas where automatic control is extensively used. Actually, water or wastewater treatment plants can be operated manually with no need for any level of automation. This may also be true for landing or departing of a single aircraft from a simple airport. When the number and sizes of the aircraft are increased, the need for automation becomes eminent. Similarly, the growing complexity of water and wastewater treatment and distribution systems, and the need to economize steadily-declining water resources, have led to the continuous and automatic monitoring of these processes. This means that the need for automation in the operation of such systems increases as the dimensions or the number of components involved in these systems increase.

It is the responsibility of water resources engineering to control and regulate water to serve a wide variety of purposes. Certain fields of water resources engineering — such as flood control, land drainage, sewerage, and highway culvert design deal primarily with the control of adverse effects of water. Water supply, irrigation, and hydroelectric power development are the three major fields which serve the utilization of water for beneficial purposes. It is basically in these fields that the largest water resources systems are developed. In recent years, pollution control or water-quality management has become an important phase of water resources engineering.

Air traffic controllers are currently engaged in a dialogue with various automated subsystems: they access databases of textual information, examine and interact with radar displays, receive and transmit flight information on hardcopy, and communicate with pilots and other controllers over audio links Many of these activities do not have the look and feel of dialogue, but convey the impression of working with a set of tools to perform a variety of tasks. Advances in computer technology over the last five to ten years reflect the increasingly successful use of this “tool metaphor,” and it is likely that future, near-term Air Traffic Control (ATC) systems will continue to reflect this approach.

The paper broadly follows the general structure of a lecture delivered at the Academie Nationale de l’Air et de l’Espace, Toulouse, France, in November 1989. The paper describes some of the work conducted by the Eurocontrol Engineering Directorate to provide automated aids to the controller and appreciably improve the efficiency of the on-line component of air traffic management and control.

Optimum performance of an automated air traffic control (ATC) system requires it to learn and modify itself to improve performance by practicing tasks over and over, solving problems, creating new terms and concepts, catching and exploiting explanations as a human air traffic controller would. Additionally, the system must check itself for errors, inconsistencies, unstable searching, identification of responsibility for decisions, and default reasoning. The multi-layer queuing model is a logic based problem solver to integrate an expert system, truth maintenance system, and queue model, to ensure efficient learning and checking are performed on-line in automated air traffic control systems. Many classes of parallel distributed processing models (also called neural network models) have been developed to address the learning issue. Expert systems have been used to organize knowledge, use rule-based and object-oriented reasoning models, and diagnose (or even execute) recommended repair plans to address problems where knowledge can be partitioned or organized with rules. Truth maintenance systems have been developed to perform indirect proofs, diagnosis, dependency-directed backtracking, dependency-directed searches, etc. to detect contradictions in conjunction with results from an expert system. The queue model and truth maintenance systems perform independently after calculations have been performed.

The Institute for Flight Guidance of the DLR (Deutsche Forschungsanstalt für Luft- und Raumfahrt), together with the BFS and air traffic controllers from Frankfurt, have developed a new air traffic planning system called COMPAS. This new system is now installed at eight controller working positions in the Regional Air Traffic Control Center of the BFS in Frankfurt.

The European Organization for the Safety of Air Navigation is an international organization set up by a number of West European states in 1962, with the original intention of carrying out en-route air traffic control over Europe. This mission was later modified, so that Eurocontrol now provides, inter alia, advice and planning facilities for other ATC authorities.

This paper briefly describes the facilities used at the Air Traffic Control Evaluation Unit (ATCEU) for assisting airspace planners and decision makers to make sure that airspace capacity can keep pace with forecast demand.

The primary objectives of air traffic control have been described as “the safe, orderly, and expeditious flow of air traffic”(see Hopkin, 1991). At a fundamental level, this represents the aided retention of object dispersal in four dimensional space-time. As the objects under air traffic control are not within the range of “humanscale”, (Hancock & Chignell, 1990) the recognition and manipulation of such objects can only be achieved via the use of technological prosthetics. Three major constraints are imposed on this general case. First, objects are limited with respect to resources; second, they are required, at some point, each to occupy essentially a common spatial location; and third, the objects themselves are under volitional control and can act independently. Separation distance covaries with object density, as does demand on communication capability and control action. The process by which such objects are controlled itself represents a record of iterative evolution to which proposed forms of automation will add an additional layering. The impact of such automation on the human operators within such a system is the major subject of this paper.

It is clear that, as Europe moves into the last decade of the century, its Air Traffic Control system is in need of major updating and re-organizing, if it is not to become the limiting factor in the projected growth in air traffic across the continent (Jackson, 1990). That a major investment in equipment and training will be required is now widely recognized, but there are important technological and systems issues to be resolved. One of the most important relates to the role of humans in ATC and the extent to which certain human roles can, and indeed should, be automated.

The need for a safe and expeditious air traffic system, along with recent and rapid developments in advanced computer-based technology, has been associated with remarkable changes in airway systems operations and in the system decision-making process. A key issue in air traffic systems concerns which human functions should be automated to maximize system performance and efficiency, without jeopardizing system safety. These decisions are quite simple for a few selected functions where human performance is superior, or in some instances inferior, to machines. The decisions become more critical in considering those functions where human and machine performance differences are subtle. These latter circumstances present the greatest challenge to the human factors professionals and system designers because of the potential human factors consequences (Adams, 1989).

The development of control systems for industrial plants have traditionally been based on mathematical models of the causal dynamic interrelationships between plant parameters (e.g., by using knowledge of the physical properties of system components). These models are then used to synthesize control algorithms on the basis of optimization techniques. However, for many important industrial plant control problems, the design problem is not only to cope with the complexity of causal dynamic interactions, but also to cope with the complexity of the end or goal-means structure of the system. Knowledge about plant goal-means structure is important in diagnosis of fault and in planning of compensatory actions (i.e., for strategic plant resource management). Furthermore, it is also important for design of human machine interfaces, as goal-means structures can be used for management, interpretation, and presentation of plant information for the operator. Models of goals-means structures are therefore important for design of integrated supervision and control systems.

Air traffic control (ATC) is one of a number of situations in which humans must use remote control. Since it is impossible for the air traffic controller to directly observe and communicate with aircraft, electronic devices must be employed to collect and represent the required information and to relay the operator’s decisions.

Throughout most of the world, but most especially in the industrialized nations, major changes are underway to modernize and update existing air traffic control systems. Not since the advent of the use of radar for air traffic control purposes following World War II have there been so many profound changes underway or on the near horizon. New high speed, large capacity computers, enhanced automation, and improved radar and communication capabilities offer us a chance to make a great step forward.

Air traffic control is a most desirable and exciting career which affords responsibility, stress, flexibility, and experience in a high technology system. Controllers must be selected, educated, trained, and licensed before they become familiar with the daily workload and sovereign in making quick decisions.

There has been much interest in recent years in the subject of workload and stress among air traffic controllers. Although some of the problems are fairly long-standing, many are becoming increasingly significant because of increasing traffic loads, the changing nature of this traffic, and technological advances in equipment. It is therefore becoming extremely important to understand the nature of the complex demands of this work and the management of those demands by controllers. Also, and perhaps most important in this type of work, we must determine whether and how the quality of system performance is affected.

Man is a day-light creature, having associated his own state of wakefulness and activity with the day-light period and, consequently, his rest and sleep state with the night-dark period, with a regular rhythm over the 24-hour span. This conditioning derives from the evolutionary adaptation of the human species to the physical environment and from the social habits of each individual. In relation to this periodic alternation, all the body functions (metabolic, respiratory, digestive, cardio-circulatory, renal, nervous) show a circadian (“circa diem” = about 24 hours) rhythm, with higher functional levels during the “ergotropic” phase (light-wake-activity) and lower levels during the “trophotropic” phase (dark-sleep-rest). This circadian rhythm is sustained by an endogenous oscillator or “body clock,” localized in the brain at the level of the suprachiasmatic nuclei, and modulated by external influences of such socio-environmental “synchronizers” as light period, timing of sleep, activity pattern, and meals. Under normal conditions of the sleep/wake cycle, the internal and external factors are in constant relation, determining a clear circadian variation of the psycho-physiological state (Minors and Waterhouse, 1986).

There are a growing number of activities in which the human operator is assigned duties that involve mental processing of information. Such processing can be experienced by the operator as being exhausting or even stressful, and such feelings were at the origin of the concept of “mental workload.” It is used as a moderator variable to explain why applications of modem technology sometimes cause such unexpected variabilities in human performance as accidents, errors, or increases in the number of health problems.

Discussion of the effects of automation in air traffic control requires an understanding of automation, the air traffic control environment, and the human operator. Without an understanding of what automation can and cannot do, it is impossible to know how it will affect the work environment. Likewise, without a clear understanding of the tasks, jobs, and goals of air traffic control, one can neither determine whether automation is appropriate nor comprehend the effects that automation will have on the work. It also requires an understanding of human abilities and limitations, preferences and biases, perceptions and cognitions. But without understanding automation and the air traffic control task, an expert in human performance can, at best, provide only vague predictions and generic warnings about how the human operator will respond to automation in air traffic control.

Several circumstances in recent years, including the destruction of Korean Airlines Flight KAL 007 (September 1983) and the shootdown of Iran Air Airbus A300 on flight IR 655 (July 1988), have resulted in a number of actions by the International Civil Aviation Organization (ICAO) to improve ICAO Standards and Recommended Practices (SARPs) in order to help prevent recurrence of such tragic incidents.

With progressive automation in air traffic control (ATC) (Hopkin, 1987, 1989a, 1989b, 1991; Stein, 1987), there is a particular need to recognize that the most effective design solutions in the automation process will be sensitive to our growing knowledge of human error. An indiscriminate application of automation technology carries with it the potential to contribute its own sources of system error. Ultimately, however, automation is intended to reduce the processing demands on the controller and to reduce the potential for human error. A better understanding of the current factors underlying the human behaviors associated with system errors will contribute directly to improved design decisions in the automation process.

As performance becomes skilled, less and less attention seems to be required to be devoted to it. As the Cambridge psychologist Sir Frederick Bartlett once perceptively remarked, “The skilled batsman has all the time in the world.” Implicitly, the more skill is gained, the more behavior is delegated to the emission of well-drilled responses, and the more attention may be paid to the anticipation of events, and the grander aspects of the overall plan. Despite the enormous gains in performance by organizing motor behavior in this way, however, if there is a trade-off between attention being paid to the larger aspects of the plan at the expense of attending to the details of motor control, the scope for minor errors in execution remaining undetected may increase.

FAA’s air traffic control mission is to promote the safe, orderly, and expeditious flow of civilian and military aircraft. Using information processed by computers and displayed on video screens at their workstations, air traffic controllers maintain the required separation between aircraft, provide safety advisories to pilots, and ensure efficient use of airspace (GAO, 1989).

Human error in computer systems is often involved in system malfunctions. Human error can be reduced by understanding seven causal factors and their implications for system performance. These factors, as described by Bailey, are: EnvironmentHuman computer interfaceOrganizational accuracy requirementsPersonal factorsSystem designTrainingWritten Instructions Of these factors, there are three over which the system designer has no control. These are personal factors, organizational accuracy requirements, and environmental factors. Since these can only be affected in a positive manner through managerial policy and subsequent enforcement, it is critical to the long-term success of the system that management pay close attention to these factors.

Successful air traffic control requires two things: the management of information and the management of consciousness. Both are important, but they are different. Automation can improve one or it can improve both. Human consciousness and human information-processing ability are limited, whereas, in practical terms, the span of attention of automation and its ability to process information is unlimited. Hence automation can greatly improve information flow to human awareness, but this can also overload it. On the other hand, if human consciousness is not part of the loop — i.e., in a completely automated system — it is quite possible for the machinery to process accurately a much larger volume of information. That is, as long as the program for the computer is correct, an important (and not necessarily correct) assumption.

It is time for tomorrow’s world...

In the following we want to look at automation and the introduction of new technology in the working sphere of air traffic controllers from a psychological perspective. The following topics will be covered: How can air traffic controllers, technology, pilots, and management be seen as a complex system?What is the meaning of flight security in this interrelation?Which factors determine human action (in our case, the actions of air traffic controllers)?What measures can increase flight security?

In developing a selection system for air traffic controllers — or any occupation, for that matter — considerations do not stop with selecting predictors. The paper presented at this conference by Della Rocca, Manning, and Wing (1991) briefly described the present political climate in which selection occurs in the United States, as well as the technical support for procedures the Federal Aviation Administration (FAA) uses to select air traffic controllers. This paper will be more theoretical, or, at least, less empirical and more thoughtful, about the concepts personnel psychologists are currently using in evaluating selection systems.

Over the next two decades, the Federal Aviation Administration’s (FAA) National Airspace System (NAS) Plan for new automated air traffic control systems will radically change the role of the Air Traffic Control Specialist (ATCS)19. While introduction of automation is not new to the occupation, the NAS Plan calls for automating more critical job tasks than could have been previously supported by technology. The challenge of identifying the characteristics of the controller to operate these future automated systems is currently facing the FAA.

This paper provides some ideas on issues related to the selection of air traffic controllers in an automated environment. It is mostly based on Saville & Holdsworth’s (SHL) work with the Civil Aviation Authority (CAA) in the UK, where we are retained to provide advice and research into the selection of controllers.

As air traffic control work is closely related to its technical environment, it has been influenced strongly by the technological developments of the last decades. Nowadays the controller benefits from digital radar-data, flight plan management, and other assistance by computers. High resolution monitors also improve working conditions in air traffic control. If we remember the usual situation of a controller in the early 60’s — sitting in a dark room looking at the flickering display of an analog radar, and writing flight strips by hand — we can appreciate the changes and improvements in today’s working conditions. This comparison might also be helpful if we consider the changes expected in the next decades: even though the development will continue in small steps towards automation, the discrepancy between today’s status and that in about another 30 years will be enormous.

Training is not an objective in itself. It is required in order to supply a sufficient number of persons, of sufficient quality, to perform specific jobs.

The Federal Aviation Administration (FAA) is responsible for the operation of the National Airspace System (NAS). The NAS is the world’s largest and most complex network of airports, airways, and air traffic control (ATC) facilities. The NAS is designed and operated to accomplish three goals: Safety of flightExpeditious movement of aircraftEfficient operation

The concerns about controller effectiveness in today’s highly automated environment suggest that new training approaches may be warranted. Just as airline pilots have had to develop new skills for dealing with cockpit automation, so now must controllers also expand their repertoire to include the skills which comprise effective resource management. Resource management training refers generally to technical training in the efficient use of system resources in the performance of a job. While air traffic controllers are not as explicitly teamoriented as flight crews, there are job parallels which suggest that performance could be improved through resource management training for skills such as leadership, planning, communication, assertiveness, situational awareness, and decision making that are often a part of flight crew resource management training. Today’s high-density and complex operations require efficient coordination among air traffic facilities and effective use of system automation. Satisfactory system performance of our newly automated air traffic systems will depend on controllers’ resource management skills.

Ask ten people to define cockpit resource management (CRM) and you will get ten different answers. There seems to be no nice, succinct way to say “the enhancement of flight safety through the optimal use of all resources and the synergistic efforts of the flight crew as a team.” Undaunted, I will now embark on the eleventh attempt to define CRM.

The introduction of advanced air traffic control systems presents a number of challenges to senior decision makers. The objective of this paper is to present an overview of six critical issues that decision makers must consider relative to training. These issues are: Method of acquisitionUse of specifications and standardsNeed for integrated work team of system engineers and training professionalsTraining deliveryCourseware validation and measurement of proficiencyConfiguration management

Mitroff (e.g., Mitroff, 1971; Mitroff & Beltz, 1972) and others have argued quite persuasively that when trying to predict an uncertain future20 one of the most effective techniques is the dialectic argument. In the dialectic approach, proponents of the two most extreme views on an issue argue the strenghts of their position and the weaknesses of the other. As a result, the strengths and weaknesses of each will be identified, and a third point-of-view, hopefully including the strengths of each and the weaknesses of neither, will be created by the observer.

The boldest challenge of the NATO Advanced Study Institute (ASI) on “Automation and Systems Issues in Air Traffic Control Systems” was to look at so-called “blue sky” solutions to future ATC systems, unprejudiced by precedent, convention, or existing concepts of ATC system evolution. Given that most ATC services are already committed to programs up to about 2005, the ATC prophet’s task is to go forward to 2040, fifty years from now. From the view point of 50 years ahead, and with a system in mind, the visionary can look back over one shoulder and identify what developments will be needed to take ATC from 2005 to 2040, if indeed there is ATC in 2040. I do not pretend to be enough of a systems expert, or an engineer in either hardware or software, to make a detailed projection; but allow me as a simple lay controller to throw my hat into the ring, as part of the debate.

This paper speculates about the evolution of air traffic control over the next fifty years or so, with particular reference to human factors aspects. While current air traffic control practices and known developments within that timescale are not ignored, these speculations do not merely extrapolate beyond them but consider some new factors that could appear in the meantime and influence the future evolution of air traffic control. It is assumed that aviation will remain a very large industry, and that fuel resources will not be a main limiting constraint.

One of the unique characteristics of an ASI is the interaction and confrontation that takes place when a group of recognized experts from diverse professions are gathered for the purpose of addressing complex societal issues. This strategy was especially effective with the particular group of participants that came together at the Hotel Villa Del Mare for the 14 days of this 1990 ASI. The iterative dynamics of the group’s formal and informal contacts facilitated a comprehensive understanding of automation and the air traffic control system that would not have been possible under any other conditions. Each of the participants made valuable contributions to the existing knowledge of the issues addressed and was rewarded, in turn, with a broader perspective of the problems and potential solutions through the integration of different views. It was this integration of unique theoretical orientations with professional experience that many of us considered the most important product of the conference. Gaining new perspectives on air traffic control problems through the cross-fertilization on many disciplines — that was the real value of the ASI!

When I started to compile material for this summary, it rapidly became obvious that the task of summarizing the ASI exemplified a point made during it: although a task can be defined and the actions required to accomplish it can be specified, it does not follow that a satisfactory task performance can be achieved or is actually feasible. The task of considering every paper, presentation, plenary discussion, group discussion, and informal meeting — in order to record every significant point, to capture the gist and enthusiasm of the discussions, and to synthesize everything into a structured and balanced summary — seemed impossible in such a short time. Good ideas were prolific, but a bald listing of them could not convey their value. Even the simplest summary of all of them would be a long document. It might also be dull, which the ASI was not. In any case, most of the ideas will appear in the published Proceedings. Therefore these remarks attempt to review the ASI from a broader perspective.

“Funny how things turn out,” mused the beekeeper, staring into his pint of still, off-the-chill bitter (flat, warm, musty) beer. “I started out as an Air Traffic Controller — until they automated the entire system. I wouldn’t like to have to do that job now.”