Download pdf of this article

Can Automation Enable a Cooperative Future ATM System?
January 31, 1997

*Cognitive Systems Engineering Laboratory, The Ohio State University, **Department of Aviation Ohio University, ***Aviation Research Laboratory University of Illinois at Urbana-Champaign

The Shape of the Future Air Traffic Management System | Air carriers have indicated a need for a less constrained, more flexible air traffic management (ATM) system. In response to this need, RTCA advanced proposals for a "free flight" system in which the airlines will be able to pursue their objectives more efficiently. This proposal assumes a higher level of automation to assist air traffic managers to detect and resolve short-term conflicts, reducing the need to impose strategic constraints on the airspace system. It also implies that human operator roles and responsibilities in the future system will be different from those in today's air traffic management system.

Figure 1 encapsulates some of the differences between the present and future system. It suggests the changes in management style that will be required., There will be a need for additional intent information, since flight path routings will normally come from pilots or airline operations centers. The controller will normally be a monitor except when short-term constraints or hazards to air traffic are detected; new technology will assist the controller, or airspace manager. Enhanced traffic displays will also be needed to aid the pilot in his increased decision-making role. It is important to note that closed-loop controller direction of air traffic is replaced by a "softer", indirect loop in which controllers must rely on pilot intent information and situation displays to "close" the control and surveillance loop.

Figure 1: Comparison of present and future ATM System Loops

Figure 2 makes clear that the construct illustrated in Figure 1 looks simple only because it does not take account of the enormous amount of information that is actually required for decision-making in the air traffic control system. A great deal of data, much of which will not be available to computers in the foreseeable future, determines what is an effective or ineffective decision in this system.

Figure 2: Information Required for Decision-Making in a Future System

Figure 3 reminds us that there is also a great deal of variability in the system, some of it under control of human managers and operators, but much of it not controllable (weather, etc.). The aviation system is highly dynamic and data describing the system changes constantly, sometimes in unanticipated ways. Coping with unexpected variance, in particular, is something computers do not do well. This fact emphasizes the systemÕs dependence on human operators who can cope with such variance, regardless of whether pilots or controllers are to be the decision-makers.

Figure 3: Sources of Variance in Air Traffic Management Data
("e" represents a source of variability or error in the process)

If decision-making is to be shared between humans and machines, and more widely distributed among controllers, traffic managers, pilots and dispatchers or AOCs, the information necessary for that decision-making must be more widely distributed than has been the case in the past. Effective management of ATM information will be absolutely critical to the success of a future system based on free flight concepts. Redistributing decision-making authority is comparatively easy: it involves only the formulation of new policies and procedures for system management. Providing real-time, relevant information to support such decision-making, in forms that support good decisions, is much harder in an information-bound system because of the enormous quantity of rapidly-changing data from which such information must be extracted. Thus, this latter challenge must be a central focus in developing such a future ATM system.

Our studies have indicated that attempts to increase operator flexibility in flight path management may often be accompanied by requirements to exchange appreciably more information among system participants. In the absence of additional information, operators find themselves unable to formulate good decision strategies and thus to effectively exercise the additional authority available to them. This was less of a problem when the system was based on management by direction; controllers had the information necessary to exercise system control, though not necessarily to exercise that control flexibly or in a manner that adequately met user needs. In a system based on more distributed control, however, the conventional system model is clearly inadequate. What is needed is a more collaborative approach to decision-making, based on more effective sharing of information among system participants. This is suggested in Figure 4, which illustrates a distributed control loop backed by an infrastructure which makes information more accessible to operators as they need it to make effective control decisions as required.

Figure 4: A Model of Shared ATM System Control and Coordination

Examples Illustrating the Need for Information Exchange | As Figure 4 indicates, there are many different individuals who need to exchange information and coordinate their activities. This information exchange is needed to support both strategic and tactical decisions.

Information Exchange to Support Strategic Decision-Making | The concept of free flight is meant to extend to strategic as well as tactical decision-making. In both cases, the goal is to give the users more flexibility in making decisions or choices based on their business concerns, subject to constraints in the system that must be considered to assure safety and efficient overall use of system capacity. One of the implications of these constraints, however, is that there will be days when a certain route or segment of that route will not be able accommodate a particular flight at all, or will not be able to accommodate that flight under enroute free flight rules, because of weather, traffic congestion or airport restrictions.

One implication is that, as the locus of control for pre-flight planning of routes is further shifted to AOCs, such restrictions need to be communicated to dispatchers so that they can consider the restrictions during their pre-flight planning. Furthermore, the dispatchers making these decisions will need flight planning tools that can take into consideration these restrictions. In essence, in a free flight environment, dispatchers need to know the intentions of the ATM system. They will need ATC forecasts just as they need weather forecasts. This view was expressed by a controller who participated in one of our studies:

Controller: "Traffic management would determine what routes would be inappropriate for free flight based on predicted weather and traffic, and would program the computer. The dispatcher would send in the flight plan 2 hours ahead of time and the computer would analyze it. Then the dispatcher would get an acknowledgment that that flight can't go today free flight. Then he could do better flight planning on it. The pilots would know before they leave that they're not flying free flight."

Such a view of the strategic portion of free flight thus raises a number of important research questions:

•  Can tools be developed to help identify such routes or route segments? (This includes communication tools to improve input from regional traffic management centers as well as tools that automatically look at collected data.)
•   Can tools be developed to help disseminate such restrictions to AOCs so that they can plan more efficiently and effectively?
•   What tools or procedures would reduce the number of situations where such restrictions would be necessary?
•   If some segments of a flight are under free flight rules while others are not, how will transitions be handled once a flight is enroute?

Tactical Decision-Making - Sharing Intent Information | Just as dispatchers need to know the intentions of the ATM system to make decisions when developing flight plans, pilots and controllers will need to know the intentions of the other aircraft in their immediate vicinity in order to decide whether a particular maneuver is acceptable. This sharing of intent information will be critical if enroute free flight is to be safely be supported with acceptably small separation distances.

This view was articulated by a number of controllers in studies that we conducted. In referring to a hypothetical scenario in which there was a loss of separation due to inadequate coordination between a flight crew and a controller, observations like the following were made:

Controller: "The overload was caused by the fact that he was surprised. Something happened that he wasn't planning on. An airplane's moving on its own without telling me about it. That starts the whole workload ball building up. What was a very manageable situation all of a sudden now gets out of hand."

Controller: "Today if this happens, I've got a surprise, I can handle it because I know what everybody else is doing. They're doing what they're supposed to be doing. But now there's a whole lot of question marks after all these call signs. In fact, that's how I would probably identify the guys who are in free flight. I'd have the computer put a question mark after his call sign because I'm not positive what he's going to be doing."

Thus, this need to share intent information relevant to tactical decision making points to a number of additional questions that must be addressed:

•  How can intent information be communicated and displayed in an efficient and effective fashion? Can intentions be adequately expressed in a form that can be "understood" by computerized conflict probe systems?
•  How will coordination be achieved to ensure timely, effective use of intent information? If a flight crew somehow broadcasts a new intention, how will its acceptability be determined (based on a first-come-first served paradigm, based some established "rules of the road", or based on discussion and negotiation between all of the flights involved)?
•  How will the controllers who are monitoring a situation determine when they need to intervene? How much detail must be provided by flights in communicating intentions, so that such decisions can be made effectively?

Maintaining Adequate Situation Awareness | Knowing the current stated intentions of other aircraft provides only part of the picture that must be considered in making tactical decisions. To be effective, the decision-maker must have full awareness of the situation (including weather, the intentions of all aircraft, the actual behaviors of all aircraft, likely changes in the intentions of all aircraft, and available contingencies). Since it is not likely that computers will be able to reason sufficiently about such complex factors, flight crews will have to have maintain this full awareness, as will the controller who is monitoring the situation. Any recommendations or warnings provided by the computer will be based on an incomplete model of the situation, and hence can be treated as only one source of information to support decision-making by the crew and the controller.

This view was reinforced by an example provided by another of the controllers in our studies:

Controller: "Some of those things, I just don't see the computer being able to predict or really help me with. You have huge thunderstorms, and every airplane in the sky is diving for a hole. I don't know how we communicate that to the computer, and how it's going to help me do the conflict prediction through this hole."

The research questions motivated by this focus include:
•   What are the circumstances where enroute free flight is feasible in the sense that flight crews and controllers can access and integrate all of the information necessary to make decisions?
•  What are the different classes of information needed? How should such information be exchanged and displayed?

Conclusions | The first theme of this paper is that, as the locus of control is shifted, access to the relevant information must also be shifted. At a strategic level, as AOCs are given increased authority to file flight plans, they will need access to much more information about likely constraints due to traffic bottlenecks and weather. Similarly, at a tactical level, while enroute, pilots and controllers will need full access to the information necessary to maintain situation awareness.

A second theme is that in some situations, decision-making requires reasoning about uncertainty and consideration of data that is not easily communicated to a computer. Both of these factors imply that people will continue to have the responsibility for maintaining full situation awareness and for integrating all of the relevant information to make decisions. Computer systems with capabilities such as conflict detection will simply be one of many sources of information that the people in the system must consider.

A third theme is that (consistent with discussions by various organizations about concepts like "dynamic density") there will be circumstances where the flexibility offered under free flight (during pre-flight planning or while enroute) will have to be curtailed. The participants in our studies generally agreed that there were situations where enroute free flight was feasible:

Controller: "There's so many times I've got one aircraft that's overtaking another and I want to say: 'Offset a couple miles and go around him or descend 300 feet and pass him.' I'll tell him the intent is to stay level. And then I'll broadcast to the other aircraft: 'He's going to come by you and all that.' Then, that takes my complete concentration away from those aircraft so that I can work on more. ... This is one way of taking some of my workload and shifting it to the cockpit."

There are also, however, situations which require levels of coordination that are difficult or impossible without more a centralized distribution of control. Thus, one of the challenges is to clearly delineate such situations, and to develop procedures for ensuring the safe, timely transition of control from AOCs to traffic managers (for strategic concerns) and from flight crews to controllers (for tactical concerns).

Finally, it is important to recognize the strategic and tactical decisions are not in actuality independent. Dispatchers and traffic managers will have to know how the impacts of their strategic choices will impact tactical decision-making. Similarly, flight crews and controllers will need a better understanding of how their tactical decisions influence the overall strategic picture. Thus, as suggested in Figure 4, dispatchers, traffic managers, flight crews and controllers will need a much better understanding of how various system components and decisions interact, and will need better real-time information to make use of this knowledge.

Acknowledgments | This work was funded by the Advanced Air Transportation Technologies program at NASA Ames Research Center under Grant Number NAG2-995. We would like to express our appreciation to Judith Orasanu and to the controllers, dispatchers, pilots and traffic managers who assisted us in completing the work.