6
Strategic Long-Range Planning

In this chapter, we provide an analysis of the human factors issues associated with four automation efforts designed to facilitate strategic air traffic control: the center TRACON automation system, the automated conflict probe, the development of four-dimensional contracts, and the surface movement advisor. As we did in the previous chapter, we analyze each piece of automation in terms of the functions performed, the context for development, and the human factors issues.

CENTER TRACON AUTOMATION SYSTEM

Functionality

The primary objective of the center TRACON automation system (CTAS) is to assist the air traffic controller in optimizing the traffic flow in the terminal area (Erzberger et al., 1993). Delays are reduced and flight paths are flown in a more economical fashion so that potential fuel savings are estimated to range from 45 to 135 kg per landing (Scott, 1994). These benefits are accomplished by providing assistance in planning and control in both routine and unexpected circumstances (e.g., changes in runway configuration). CTAS is also capable of providing advice to controllers regarding particular airline preferences. CTAS is comprised of three separate tools or elements, each supporting different classes of air traffic control personnel, located in different facilities, and coordinating different phases of the approach (Figure 6.1):



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation 6 Strategic Long-Range Planning In this chapter, we provide an analysis of the human factors issues associated with four automation efforts designed to facilitate strategic air traffic control: the center TRACON automation system, the automated conflict probe, the development of four-dimensional contracts, and the surface movement advisor. As we did in the previous chapter, we analyze each piece of automation in terms of the functions performed, the context for development, and the human factors issues. CENTER TRACON AUTOMATION SYSTEM Functionality The primary objective of the center TRACON automation system (CTAS) is to assist the air traffic controller in optimizing the traffic flow in the terminal area (Erzberger et al., 1993). Delays are reduced and flight paths are flown in a more economical fashion so that potential fuel savings are estimated to range from 45 to 135 kg per landing (Scott, 1994). These benefits are accomplished by providing assistance in planning and control in both routine and unexpected circumstances (e.g., changes in runway configuration). CTAS is also capable of providing advice to controllers regarding particular airline preferences. CTAS is comprised of three separate tools or elements, each supporting different classes of air traffic control personnel, located in different facilities, and coordinating different phases of the approach (Figure 6.1):

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation FIGURE 6.1 Elements of the center TRACON automation system (CTAS). Source: National Aeronautics and Space Administration. The traffic management advisor (TMA) supports the TRACON and en route traffic management controllers, primarily in developing an optimal plan, to assign each aircraft a scheduled time of arrival at a downstream point, like a final approach fix or runway threshold, and a sequence of arrival, relative to other aircraft approaching the terminal area. The traffic management advisor begins to compute these for inbound aircraft at a point about 200 miles or 45 minutes from the final approach. The plan is designed to optimize the overall flow of the set of aircraft, as well as the fuel consumption of each individual aircraft. At the same time, it accounts for various constraints on runway availability and aircraft maneuverability. The plan is also accompanied by an assessment of flight path changes to be implemented in order to accomplish the plan. A set of three displays assists the traffic management coordinator in evaluating the plan (Figure 6.2, see color plate). These include a time line of scheduled and estimated times of arrivals for the aircraft, a listing of alternative runway configurations, and a load graph, which indicates the anticipated traffic load across designated points in the airspace in 15-minute increments. The displays can be presented in large-screen formats for group viewing (Figure 6.3, see color plate). The actual implementation of the plan generated by the traffic management coordinator with the assistance of the traffic management advisor is carried out by the other two elements of CTAS, the descent advisor, and the final approach spacing tool. The descent advisor (DA) provides controllers at the final sector of the en

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation route center with advice on proper speed, altitude, and (occasionally) heading control necessary to accomplish the plan generated by the traffic management advisor. The critical algorithm underlying the descent advisor is a four-dimensional predictor that is individually tailored for each aircraft, based on that aircraft's type and preferred maneuver, along with local atmospheric data. This predictor generates a set of possible trajectories for the aircraft to implement the traffic management advisor plan. The descent advisor then provides the controller with a set of advisories regarding speed, top of descent point, and descent speed. In cases in which these parameters are not sufficient to accomplish the plan, path stretching advisories are offered that advise lateral maneuvers. The descent advisor also contains a conflict probe that will monitor for possible conflicts up to 20 minutes ahead. If such conflicts are detected, it will offer resolution advisories, based initially on speed and altitude changes. If none of these is feasible, lateral maneuvers will be offered as a solution. Figure 6.4 (see color plate) illustrates the descent advisor display. Controllers read the advice of the descent advisor on the fourth line of the data tag for each aircraft. In addition, markers on the plan view displays indicate the location at which flight path trajectory changes should be issued. Time lines, similar to those provided by the traffic management advisor, are also available at the side of the display. The final approach spacing tool (FAST) is the corresponding advisory tool designed to support the TRACON controller in implementing the traffic management advisor plan, by issuing speed and heading advisories and runway assignments necessary to maintain optimal spacing between aircraft of different classes (Davis et al., 1994; Lee and Davis, 1995). An important secondary function of the final approach spacing tool is its ability to rapidly adjust to—and reschedule on the basis of—unexpected events like a missed approach or a sudden unexpected runway closure. Like the descent advisor, the controller receives advice in the fourth line of the data tag, and also has access to time lines (Figure 6.5, see color plate). The final approach spacing tool exists in two versions: the passive FAST provides only aircraft sequence and runway assignments, and the active FAST includes speed and heading advisories. A system with similar functions, known as COMPAS (computer-oriented metering planning and advisory system), was developed by the German Aerospace Research Establishment and has been operational in Frankfurt since 1991 (Völckers, 1991). The system attempts to assist the controller in the planning and control of approach traffic. Based on flight plans and radar data, the system calculates the arrival time of each aircraft, taking into consideration such parameters as aircraft performance, traffic proximity, and wake vortex separation minima. On the basis of these calculated arrival times, the system establishes a landing sequence, as well as a nominal gate arrival time for each aircraft. The difference between the calculated time and the nominal time is then transformed

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation into a medium-term plan by which traffic flow can be smoothed, starting outside the terminal area. The goal of this plan is to reduce aggregate delay across the entire traffic stream. The interface for the COMPAS presents the controller with a sequencing time line, with arrivals ordered from latest to earliest, top to bottom. The time associated with each aircraft represents the estimated arrival time, over either the metering fix (in the case of the en route controller) or the approach gate (for the approach controller). Aircraft weight class and approach direction are represented in the display. The control advisories themselves are presented as one of four possible characters beside the aircraft label: X to expedite up to two minutes, O for no action, R for delay up to four minutes, and H to hold for more than four minutes. Operational experience with COMPAS has demonstrated reductions in planning and coordination workload, as well as reductions in the time spent on coordination and smoothing of the traffic flow in the terminal area. Notice that COMPAS provides general resolution advisories (e.g., ''hold for more than four minutes"); the descent advisor of CTAS provides another level of assistance—namely, the specific action by which a conflict should be resolved (e.g., "descend to flight-level 70"). History The main impetus toward the development of CTAS has been the loss of capacity in airport arrival and landings. Limitations in prediction of trajectories and weather have led to spaces on the final approaches that are not occupied by an aircraft, thus creating delays or not meeting the actual capabilities of an airport's true capacity. In the 1980s, the National Aeronautics and Space Administration (NASA) and the Federal Aviation Administration Technical Center began an in-house research and development project to develop the software tools for achieving this optimization (Erzberger and Tobias, 1986), working closely with controllers and human factors professionals to create a fielded system. During the mid-1990s, this system has received several field tests at Dallas-Fort Worth International Airport and at the Denver airport and center. It is also being installed at Schiphol Airport in Amsterdam, the Netherlands. Human Factors Implementation Human factors has played a relatively important role in the maturation of CTAS, from concept, to laboratory prototype, to simulation, to field test (Erzberger and Tobias, 1986; Tobias et al., 1989; Harwood et al., in press). From 1992 to 1997, approximately 30,000 person hours of human factors expertise have been devoted to CTAS development and fielding. In part, the successful implementation of the human factors input was a result of the fact that the development

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation took place at NASA laboratories, with ready access to human factors professionals and active participation of controllers in developing the specifications. The development was not under constraints related to contract delivery time or required specifications. Human factors implementation was also facilitated in part by the frequent input of controllers to the design concepts of functions at all phases and frequent human-in-the-loop evaluations at varying levels of simulation fidelity. The controller's input was filtered by human factors professionals (Lee and Davis, 1995; Harwood et al., in press). Another important factor is that these evaluations (and system changes based thereon) have continued as the system is being filed tested at the Dallas and Denver facilities (Harwood et al., in press). In particular, developers realized the need for extensive input from a team of controllers at the facility, in order to tailor the system to facility-specific characteristics. The introduction process was quite time-consuming, taking place over several years. This proved necessary (and advantageous), both in order to secure inputs from controllers at all levels, and also in order for human factors professionals and engineers on the design team to thoroughly familiarize themselves with the culture and operating procedures at the Denver and Dallas-Fort Worth facilities; this, in turn, was necessary in order for the trust of the operational controllers to be gained and for the CTAS advisories to be employed successfully. It is also important to note that the system was designed to have a minimal effect on the existing automated systems (HOST and the automated radar terminal system, ARTS) and on existing procedures. Finally, it should be stressed that CTAS is presented to controllers with the philosophy that it is an advisory aid, designed to improve their capabilities, rather than as an automation replacement. That is, nothing in CTAS qualitatively alters the way in which controllers implement their control over the aircraft. Human Factors Issues Cognitive Task Analysis A cognitive task analysis reveals that CTAS supports the controller's task in three critical respects, addressing the vulnerabilities identified in the panel's Phase I report. First, its four-dimensional predictive capabilities compensate for difficulties that the unaided controller will have in predicting and visualizing the long-term (i.e., five minutes) implications of multiple, complex, speed-varying trajectories subjected to various constraints, such as fuel consumption, winds, and runway configuration. With the current system, these limits of the unaided human constrain the flexibility of considering a variety of traffic plans. Second, its interactive planning and scheduling capabilities allow multiple solutions to be evaluated off-line, with the graphics feedback available in the time lines, to facilitate the choice of plans. Here also the system supports the workload-intensive

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation aspects of planning (Johannsen and Rouse, 1983; Tulga and Sheridan, 1980), particularly prevalent when multiple plans need to be compared. Finally, CTAS, particularly the final approach spacing tool, supports the controller's ability to deal with the high-workload imposed by unexpected and complex events, characterized for example by a missed approach or an unanticipated runway closure. The first and second of these tasks primarily affect the efficiency of system performance, whereas the latter appears to have direct and beneficial safety implications. Workload A stated objective of CTAS is that it will not increase controller workload; indeed, field tests of the system reveal that this criterion has been met (Harwood et al., in press). As noted above, CTAS has the potential to reduce workload during the "spikes" imposed by unexpected scheduling and spacing requirements due to a missed approach or closed runway. However, it is also the case that workload may be shifted somewhat with the introduction of CTAS. Relying on an added channel of display information, rather than the controller's own mental judgment, may impose an increase in visual workload. In fact, any new set of procedures (such as those associated with CTAS) would be likely to impose some transient workload increase. Finally, although not yet reported, a tool such as CTAS does have the potential of advising maneuvers that create an airspace considerably more complex than that viewed under unaided conditions (Wyndemere, 1996). In such a case, controller monitoring and perceptual workload may be increased by the controller's effort to maintain a full level of situation awareness of the more complex airspace, an issue that we revisit later in this chapter and in Chapter 10. Training The general approach to CTAS is to first provide simulation, then provide a shadowing of the real traffic off-line in the system. In the shadowing mode, CTAS elements provide the advice, and the controller can compare clearances that he or she might provide on the basis of that advice with clearances more typical of an unadvised controller and evaluate the differences (Lee and Davis, 1995). The controller can then determine the rationale behind the automated advisory. This builds confidence that the computer can provide advice to maintain separation. One might anticipate the need for some training of pilots regarding the CTAS system, not because procedures are altered, but because the nature of the clearances and instructions may be changed, relative to the more standardized, space-based approaches (i.e., using the standard terminal arrival system) in a non-CTAS facility.

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation Communication and Coordination Because of the philosophy by which the traffic management advisor plans are implemented via the descent advisor and the final approach spacing tool advisories, CTAS imposes a relatively heavy communication load between operators and facilities. This is supported via digital data transfer rather than voice communications. Furthermore, the philosophy of repeated displays across different environments (e.g., the time line seen in Figure 6.2) supports greater communications and coordination between operators, in that these can better support a shared situation awareness of the implications of different schedules. The extent to which ground-air communications are altered by CTAS remains unclear. At least one field study of the final approach spacing tool (Harwood et al., 1997), carried out at the Dallas Airport over a 6-month period indicated that the system imposed no increase in overall communications, although the nature of the communications was altered somewhat, involving more messages pertaining to runway assignments and sequencing. Automation Issues CTAS is sufficiently recent in its introduction that there has not been time to identify specific human factors automation issues on the basis of operational experience (e.g., operational errors or aviation safety reporting system incidents). However, analysis of system capabilities suggests at least some of these that may surface. Mode Errors CTAS does contain some multimode operations. For example, with the descent advisor, controllers can choose a route intercept or a waypoint capture mode for individual aircraft, as well as one of three possible speed control modes for all aircraft (Erzberger and Nedell, 1989). However, the system appears to be designed so that different modes are prominently displayed, and active decisions must be carried out to change modes, so that mode errors would appear to be very unlikely. Mistrust There would appear to be a real possibility that the advice offered by CTAS could be initially mistrusted by controllers if it differed substantially from the way in which control is typically accomplished. It would seem that such trust must be carefully built through careful training with both simulated and live traffic. Indeed, Harwood et al. (in press) noted the increase in controller confidence after they had used the system (and relied on the final approach spacing tool advice) with live traffic. This provided the opportunity to see the real improvement in traffic flow (13 percent) that was achieved.

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation Overtrust and Complacency Currently, the philosophy of system implementation safeguards against undue complacency. This is because controllers must still give the actual clearances orally, as they would in a nonaided situation. Hence, they remain more likely to actively think about those clearances, for example, than they would in a system in which clearances could be relayed via data link with a simple keystroke (i.e., automation of response execution in Chapter 1). Complacency is not generally recognized as a concern until an incident of automation failure occurs, in which the human's failure to intervene or resume control appropriately is attributed to such complacency. No such incidents have been observed with CTAS. The advice-giving algorithms were thoroughly tested, and in operational trials have yet to fail; alternatively, if inappropriate advice was ever provided, controllers were sufficiently noncomplacent that they chose to ignore it. In short, the system has been in use for an insufficient time for trust to reach the possible excessive level at which it could be described as complacency. Past experience with other systems indicates that systems can fail, in ways that cannot be foreseen in advance (e.g., the software does not anticipate a particular unusual circumstance). Furthermore, despite the design philosophy that appears to keep the controller a relatively active participant in the control loop, it is also the case that the primary objective of CTAS is to increase the efficiency (and therefore saturation) of the terminal airspace. Such a circumstance would make recovery more difficult, should problems emerge for which CTAS would be unable to offer reliable advice. Skill Degradation As with complacency, so with skill degradation: CTAS has not been used long enough to determine whether this is an issue. Yet it is easy to imagine circumstances in which controllers increasingly begin to rely on CTAS advice, relaying this as instructions to pilots, losing the skills at selecting maneuvers on their own. This may be more problematic still, to the extent that the maneuvers recommended by CTAS are qualitatively different from those that would previously have been issued by unaided controllers. At this time, a clear tabulation of maneuver differences with and without CTAS has not been carried out. Organization The organizational implications of CTAS remain uncertain. A strength of the system is that it is designed to be advisory only; by not directly affecting

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation required procedures, the negative impact on organizational functioning should be minimized. However, it is possible that subtle shifts in authority from the R-side controller to the D-side (who is more likely to have direct access to CTAS advisories) could have unpredictable consequences. We explore these consequences further in the discussion of conflict probes in the following section. Conclusion CTAS appears to be a well-conceived automation concept, addressing a valid concern of the less automated system and designed with an appropriate philosophy that is based on automated advice-giving, rather than automation-based control. As such it is characterized by a relatively low point on the level of automation action scale, discussed in Chapter 1, which accordingly diminishes (but does not eliminate) the extent of concern for complacency. Finally, CTAS has been developed and introduced gradually, in a manner sensitive to human factors issues, and to the importance of filtered controller input into the functioning of the system. Careful human factors monitoring of the system's field use should be continued. CONFLICT PROBE AND INTERACTIVE PLANNING The core of the controller's job is to maintain a continuous flow of air traffic while also preserving adequate separation. There are three interrelated automation functionalities that can potentially assist in these goals: conflict probes, interactive planning tools, and conflict resolution advisors. The conflict probe is essentially a preview of the current flight trajectory of a given aircraft, to assess whether it will create a loss of separation with another aircraft at some time in the future. Current probes exist in the ARTS and HOST computer systems, yielding alerts if conflicts are predicted (discussed in Chapter 4). Similar conflict probe logic also characterizes the TCAS system (discussed in Chapter 5). These current air traffic control probes are not sophisticated, in the sense that their predictive logic is based on an extrapolation of the current ground velocity (or, in the case of TCAS, the rate of closure). They may also be described as tactical, in that they forecast only a short duration (i.e., a few minutes or less) into the future. In contrast, however, far more intelligent probes, such as those embedded in CTAS, can include models of different aircraft capabilities, head winds, and even flight plans, to more accurately estimate the future four-dimensional trajectory of the aircraft. Smart probes, such as those incorporated into CTAS and COMPAS are far more strategic in nature, allowing much longer look-ahead. Figure 6.6 illustrates a CTAS conflict probe display (see color plate). It should be noted, however, that, although systems such as COMPAS and CTAS are highly sophisticated conflict resolution systems, they leave the final authority for implementing that resolution squarely with the human. Some other development efforts

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation over the years (e.g., the U.S. AERA system, the European ARC2000 developed by Eurocontrol) have not always embraced the same approach and have investigated the potential for fully automated conflict resolution. The ARC2000 program, for example, sought to develop a fully automated strategic resolution system. Although the development of efficient conflict resolution algorithms proved somewhat difficult, the ARC2000 is directly credited with the development of later PHARE tools (such as the PHARE HIPS, described below). Any conflict probe is by definition based on a prediction of future behavior of the aircraft involved. Such prediction or intent inferencing must of necessity be imperfect, and it will be more so, the farther into the future that behavior is predicted. Hence, a conflict probe should be able to differentiate most likely scenarios from worst-case scenarios, the former being defined by the best guess on future behavior, and the latter being defined by the margins of uncertainty if the two aircraft maneuver toward a conflict. This uncertainty can either be portrayed graphically and continuously over time (Figure 6.7), or discretely, at a given time horizon (often selected as 20 minutes). Once a conflict is probed and identified, it must then be negotiated. Automation has the capability of providing two further services to assist with this negotiation. Computers can recommend a course of action to resolve the conflict (automated conflict resolution), or they can provide interactive graphical tools as a decision aid, to assist controllers in developing a solution themselves to resolve the conflict. Using the framework presented at the beginning of Chapter 1, we note that conflict resolution is a higher level of automation than tool-based decision aiding. Ironically, however, air traffic control has proceeded more directly to implementing conflict resolution than to providing interactive tools for decision aiding. For example, we note that CTAS and COMPAS both employ automation to formulate recommended solutions for the controller to either accept (and implement through traditional procedures) or reject, and both are in active service at certain airports (Frankfurt, Denver, Dallas-Fort Worth). A less mature level of development characterizes interactive decision aids, two of which we describe in some detail below: the user request evaluation tool developed for use in the United States and the highly interactive problem solver (HIPS) developed by Eurocontrol. User Request Evaluation Tool Functionality The user request evaluation tool (URET), developed by the MITRE Corporation for assistance to the en route controller, provides a conflict probe based on a 20-minute look-ahead capability (Brudnick et al., 1996). The probe is a "smart" one, accounting for different flight plans, aircraft models (i.e., flight capabilities), and anticipated head winds, in determining the best estimate of each aircraft's

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation FIGURE 6.7 Margin of uncertainty in conflict probe prediction. position 20 minutes into the future. (It does not, however, take into account any future flight plans based on scheduled mode changes in a flight management system). The results of the probe are displayed to the D-side controller in two modes: a graphic mode portrays the flight path on a large-scale electronic map (Figure 6.8, see color plate). Projected flight paths are color-coded: a red code indicates a likely future conflict, and an amber code indicates a possible conflict (Figure 6.9, see color plate). The latter represents a wider bound of uncertainty of future behavior. A tabular mode, visible concurrently, characterizes each aircraft by a line of text portraying flight information; it will also color-code any pair of aircraft involved in a predicted conflict (Figure 6.10, see color plate). The key interactive feature of the user request evaluation tool is the planning mode, which allows the D-side controller to play what-if scenarios by graphically examining the implications of alternative instructions that could be given to one or both aircraft. Thus, for example, a controller might see that a current predicted conflict can be eliminated by increasing the air speed of one of the aircraft by 30 knots. The recommended change can then be suggested to the R-side controller, who may then choose to implement the instructions. Clear and salient indications

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation Failure Recovery The prime issues for failure recovery are twofold, both emanating from potential problems identified above. First, it is possible to envision a scenario in which the deployment of interactive tools has enabled more complex (and possibly more densely packed) traffic flow. Second, this scenario situation has also left the R-side controller with reduced situation awareness of the current airspace (because trajectory changes were not imposed by his own decisions). A sudden failure of the user request evaluation tool or conflict probe system could thereby leave the R-side controller more vulnerable in issuing the rapid tactical commands necessary to avoid conflict situations. User Acceptance As with many sophisticated forms of automation, if systems like these are not carefully designed and introduced with adequate concern for controller training, the potential exists for limited user acceptance to threaten job satisfaction, which may in turn be reflected in perceived job insecurity. The more capable such automated systems are, the more likely such fears become. Furthermore, history suggests that such fears are not always unwarranted; in 1982, the FAA's modernization plans were presented to the U.S. Congress with the promise that they would reduce future staffing requirements (Stix, 1994). Some also fear that advanced air traffic control automation, if not well-designed, may erode the job satisfaction a controller derives from resolving a challenging situation (Harwood, 1993). Conclusion Interactive planning tools appear to offer many of the benefits of automated information collection and integration (providing more easily visualizable predictive information to support human problem solving), without inviting some of the most obvious costs associated with automation of response (complacency and skill degradation). Nevertheless, predicted effects, discussed above, remain uncertain and should be the focus of continued evaluation. FOUR-DIMENSIONAL CONTRACTS Air traffic management seeks to solve a four-dimensional space-time problem. Aircraft flight paths in three dimensions of space (latitude, longitude, altitude) must be coordinated over time so as to be conflict-free. Currently, controllers and flow managers solve this four-dimensional problem by forming a mental picture of aircraft trajectories in the future. The picture of the future traffic pattern is updated periodically as new data about aircraft positions and weather

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation are obtained. Moreover, current procedures dictate that controllers have control over the aircraft's flight path, so that they can anticipate potential conflicts and plan for the future. The controller also has available display tools for short-term projection of flight paths. Thus, the controller's cognitive skills in planning and the spatiotemporal projection of flight paths, combined with display aids, form the basis for the current system of air traffic management. Although experienced controllers have developed considerable cognitive skill in trajectory prediction, additional tools may be necessary to facilitate this skill under high-workload and as traffic density increases. Such tools have also been proposed because current air traffic management is thought to be less than optimal. There is a disparity between the accurate, fuel-efficient trajectory that an aircraft fitted with a flight management system (FMS) can fly and the more limited, constrained flight path that air traffic management can offer. As a result, the benefits offered by the FMS cannot be realized. One solution to this problem is to down-link FMS-derived information on the current and future aircraft trajectory to ground-air traffic management systems. Automated tools could then be developed to help the controller in using this information to negotiate with the pilot a flight trajectory more compatible with the FMS-derived path and to detect and resolve conflicts over longer periods of time. In particular, automated tools have been proposed to improve capacity through more accurate navigation in four-dimensions. Recent work in Europe has been aimed at developing such four-dimensional tools. Functionality The Programme for Harmonised Air Traffic Management Research in Eurocontrol (PHARE) has specified a medium-term future air traffic management scenario that comprises (among other things) a suite of tools for pilots and controllers aimed at facilitating trajectory prediction and conflict detection. The medium-term scenario envisions a process of negotiation between airborne and ground-based systems, whereby an agreed flight path can be flown with minimal ground intervention. The resultant trajectory for a given aircraft could then be represented as a four-dimensional "tube" through space (Eurocontrol, 1996). Each aircraft would be assigned a tube, resulting in a number of tubes representing all the traffic in a given airspace (Figure 6.12). The four-dimensional tube can be represented as a three-dimensional "bubble" that moves through space such that its position and size are specified functions of time (see Figure 6.13). The precise cross-sectional dimensions of such a tube would vary dynamically with such factors as traffic density and weather disturbances. The tube may grow or shrink asymmetrically along any of the three possible dimensions of space, but it will always remain aligned with respect to the anticipated route. An aircraft would be required to remain within

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation FIGURE 6.12 Four-dimensional (4-D) contract concept. Source: Eurocontrol (1996). The copyright vests in the European Organisation for the Safety of Air Navigation (EUROCONTROL); the CENA (Centre d'études de la navigation aérienne); the STNA (Service technique de la navigation aérienne); the NLR (Nationaal Lucht- en Ruimtevaartlaboratorium); the RLD (Rijksluchtvaartdienst); the LVB (Luchtverkeersbeveiliging); the DLR (Deutsche Forschungsanstalt für Luft- und Raumfahrt); the DFS (Deutsche Flugsicherung GmbH); the UK CAA (Civil Aviation Authority); the NATS (National Air Traffic Services), and the DRA (Defence Research Agency). the tube at all times. In fact, the tube is the basis for a negotiated "contract" between the pilot and the controller; hence the term 4-D contract. Negotiation of the contract will necessarily involve heavy use of data link to support air-ground communication. In an experimental flight management system concept being evaluated by Eurocontrol, aircraft intentions (a four-dimensional trajectory derived by the system) would be down-linked to air traffic management, who would then up-link their requirements in terms of route or time constraints. A prediction system would then calculate the detailed four-dimensional trajectory in a manner that meets the system's specifications within the air traffic management constraints. This would then be down-linked to air traffic management, who will then have to approve the trajectory tube. In principle, the entire trajectory tube, from origin to destination, would be specified and agreed on as the basis for the contract. To reduce data link overhead, default tube parameters could be used for different flight segments, so that only limited tube reference information would need to be up-linked (Eurocontrol, 1996). In the 4-D contract scenario, the pilot will be free to modify the flight path within the tube (Flohr, 1997). For example, the pilot may deviate laterally at will

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation FIGURE 6.13 Lateral route and tube for four-dimensional (4-D) contract. Source: Eurocontrol (1996). The copyright vests in the European Organisation for the Safety of Air Navigation (EUROCONTROL); the CENA (Centre d'études de la navigation aérienne); the STNA (Service technique de la navigation aérienne); the NLR (Nationaal Lucht- en Ruimtevaartlaboratorium); the RLD (Rijksluchtvaartdienst); the LVB (Luchtverkeersbeveiliging); the DLR (Deutsche Forschungsanstalt für Luft- und Raumfahrt); the DFS (Deutsche Flugsicherung GmbH); the UK CAA (Civil Aviation Authority); the NATS (National Air Traffic Services), and the DRA (Defence Research Agency). so long as the left and right tube boundaries are not breached (see Figure 6.13). In this respect, the 4-D contract concept is similar to the U.S. air traffic management concept of free flight, albeit in a more limited form: the degree of pilot freedom lies somewhere between current practices and "advanced" free flight, in which aircraft have much greater flexibility in setting and changing flight paths (RTCA, 1995b). Human Factors Issues The 4-D contract concept is relatively new, and validations of the concept and human factors studies (simulations and field trials) are still ongoing. Two demonstration projects have been completed to date: PHARE Demonstration 1 and 2, PD/1 and PD/2. PD/1 examined, among other issues, acceptance by controllers of a sector-based 4-D contract system for en route air traffic management. As mentioned earlier, initial analyses suggest that such a system can reduce subjective tactical workload without any cost in user acceptance (Schroter, 1996; National Air Traffic Services Limited, 1996).

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation An important technical and procedural issue in the 4-D contract concept is whether the contract applies locally (e.g., to a sector) or globally (to multiple sectors). The original concept envisages a negotiated contract from origin to destination, gate to gate. This is clearly what the pilot and the airlines would prefer, because it would be consistent with the capabilities that the FMS provides and would facilitate pilot flight planning. Controllers, however, may prefer to negotiate contracts sector by sector, because this would give them greater flexibility in management of the traffic pattern, particularly in response to unexpected events, weather disturbances, etc. Of course, whether or not 4-D contracts are negotiated within or across sectors, the controller will still be responsible for the separation of aircraft. Also, controllers will be able to cancel a contract in response to unanticipated conditions at any time. When this is done, the aircraft will be under tactical control from the ground, as in current practice. Once the condition has passed, however, a new contract can be negotiated. A significant human factors concern is whether such negotiations can be undertaken efficiently and safely in a time-critical environment. Procedures for unambiguous and uninterruptable trajectory or clearance negotiation must be worked out. Thus, 4-D contracts will change aspects of the controller's job, but will not fundamentally alter responsibility for separation. The question is, will the changes affect the ability of controllers to maintain separation? Only limited data are available to answer this question. On one hand, limited, routine clearances that are communicated to the pilot under the current system may be eliminated, so that controllers may be able to devote greater time to longer-range conflict prediction and planning. On the other hand, the current system is one in which the controller knows that the aircraft will follow a precise, specified path. This will be replaced by a system in which there will be some uncertainty about the aircraft's future position. The larger the 4-D tube, the greater the uncertainty. It is difficult to predict how controllers will react to such a system. One possibility is that they will attempt to reduce the uncertainty by querying pilots more frequently, which would tend to increase communications workload. Endsley (1996b) reported such an effect in an experimental evaluation of a free flight scenario. However, it is also possible that controllers will adapt to the system and attempt to calibrate their level of uncertainty in advance by negotiating narrower tubes for anticipated problem areas within a sector and larger tubes elsewhere. The development of the 4-D contract concept has been accompanied by attention to human factors enhancements, particularly in the controller's tools and operating procedures. One product of the PHARE effort will be a set of integrated controller tools, known as the PHARE advanced tools, that incorporates the following capabilities:

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation The trajectory predictor predicts the onward path of aircraft in four-dimensions. The conflict probe predicts conflicts based on the output of the trajectory predictor. The flight path monitor detects deviations from planned flight trajectories. The negotiation manager processes communication (air-ground and ground-ground) to facilitate flight path negotiation. The problem solver proposes solutions to resolve conflicts (as predicted by the conflict probe) or other problems. The arrival manager provides scheduling/sequencing information for arrival traffic within the terminal maneuvering area. The departure manager provides departure advisories to optimize flow into the en route sector. The cooperative tools manage controller workload by monitoring, predicting, and adapting to future task demands. The tactical load smoother creates interface enhancements aimed at improving high level, multisector flow planning. Another important human factors issue concerns how controllers will interact with displays of the 4-D contracts. Given the graphical and spatial qualities of the four-dimensional concept, it would seem appropriate to make controller interaction with the display also graphical and spatial rather than alphanumeric. This is consistent with a direct-manipulation approach to human-computer interaction (Norman, 1993; Robertson et al., 1993). The PHARE advanced tools incorporate the highly interactive problem solver (discussed in the previous section) that permits the controller to resolve traffic conflicts by interacting directly with graphical depictions of tubes in the sky for a given traffic sample. Dynamic data from underlying databases (with respect, for example, to weather and aircraft performance) are transformed and integrated into the displayed flight navigation tubes. Initial trials with active controllers have suggested that this approach can provide substantial benefits in conflict resolution time (and hence traffic throughput), as well as high levels of user acceptance (National Air Traffic Services Limited, 1996). Failure Recovery When the system has saturated the airspace and a partial failure occurs, the use of 4-D contracting tools, like CTAS and interactive conflict resolution tools, may lead to problems in achieving effective and timely recovery.

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation AUTOMATED SUPPORT FOR AIRPORT OPERATIONS: THE SURFACE MOVEMENT ADVISOR PROGRAM Functionality Increased automation has been viewed by the FAA as a means of improving the efficiency of airport operations while maintaining safe taxiway navigation, takeoffs, and landings, especially during low-visibility operations. Delays in air traffic translate to extensive costs for airlines and passengers. In response to projected increases in air traffic, the U.S. aviation industry and the FAA are investing billions of dollars to increase airport capacity. However, capacity increases must be supported with improvements in the ability of the national airspace system to take advantage of capacity. Airport operations are a significant candidate for improvement (Jones and Young, 1996). To address efficiency concerns, the FAA, in collaboration with NASA, is undertaking large-scale development activities to provide controllers, pilots, airfield managers, and airline operations personnel with cues that enhance situation awareness and with automated support of surface traffic planning. The surface movement advisor project, a joint activity of the FAA and NASA, is being developed to improve the efficiency with which airport facilities operate. The advisor, which is in the concept development and demonstration phase and is undergoing prototype testing at the Atlanta airport, would integrate information from and share information among FAA controllers and air traffic control supervisors, FAA traffic management coordinators (as well as the ATCSCC central flow control facility), ramp operators, airport managers, airline operators, and pilots. The surface movement advisor architectural concept, illustrated in Figure 6.14, is based on a server that collects the following data: information from the FAA tower (e.g., runway configurations), surveillance data (e.g., radar data), weather, real-time aircraft status updates, gate information, airline schedules, and flight plans. The advisor includes automated analysis, prediction, and planning tools (i.e., performance histograms, prediction algorithms, airport operations procedure aids, and statistical analyzers) and distributes, as appropriate, collected data as well as analyses, predictions, and plans to FAA, airport, and airline personnel. This information will assist cooperating personnel to optimize gate resource utilization; balance taxi departure loads; improve gate scheduling and rescheduling; facilitate airport operations analysis; improve crew scheduling; and reduce voice radio traffic (National Aeronautics and Space Administration, 1996). Potential long-term upgrades to the surface movement advisor include performance improvements based on actual customer use and feedback; integration of air traffic management technologies such as CTAS; implementation of data warehousing and data mining of online airport traffic data (e.g., analysis of cause

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation FIGURE 6.14 Surface movement advisor (SMA) system overview. Source: Federal Aviation Administration. and effect relationships between data sources such as weather, operations, and schedules); implementation of wireless mobile computing technologies to promote wider surface movement advisor data access; and integration of the advisor with surface traffic development and test facilities (National Aeronautics and Space Administration, 1996). Human Factors Implementation The surface movement advisor is being developed according to the "build a little, test a little" development philosophy that includes early involvement of users and human factors professionals and ongoing evaluations using mixed methodologies (see Chapter 9 for a discussion of the advantages of these activities). Its subsystems and features are developed with the support of a surface development and test facility sponsored jointly by NASA and the FAA. The facility supports prototyping and simulation studies that involve test designs developed by human factors professionals and the participation of air traffic controllers, flight data and clearance delivery personnel, traffic management coordinators, tower cab coordinators, supervisors, ramp controllers, "pseudo-pilots," and airport

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation operators. It provides a real-time, interactive, simulated operational airport environment, and its studies support validation of designs as well as development of site-specific adaptations for the surface movement advisor (National Aeronautics and Space Administration, 1997d). The facility is therefore both a research and a development tool. In particular, it holds great promise as a testbed for evaluating the interactive effects of introducing multiple automation features into the extant system over time. Human Factors Issues Trust As this report notes repeatedly with respect to conflict avoidance in ground operations, one of the greatest causes of mistrust or misapplied trust results when controllers or pilots fail to develop mental models appropriate to the system and task at hand. An appropriate mental model may be considered prerequisite for situation awareness. Both controllers and pilots, as well as airport managers and airline operations personnel, will be expected to develop mental models and situation awareness pertinent to the efficiency of airport operations (e.g., awareness of schedules, gate availabilities, and clearances). Mental Models, Situation Awareness, and Loss of Skill The risks of operator inability to develop and apply a mental model of the system's activity, operator inability to monitor fast-paced machine actions, and associated loss of situation awareness are introduced when system automation (e.g., improved surveillance accuracy coupled with sophisticated airport movement area safety system logic, as well as automated planners and schedulers that advise high-efficiency operations relying upon the automated surveillance and processing technology) permits more complex activities (e.g., the movement of greater numbers of aircraft, including under low-visibility conditions). The loss of situation awareness may be accompanied by degradation of skills, if the operator has not maintained proficiency in tasks that are normally performed by the automation. For the surface movement advisor, such tasks may include monitoring the positions and movements of aircraft on the ground (if the surface movement advisor introduces substantial automation to this task), scheduling clearances, coordinating flight plans, and assessing capacity and use of airport resources. The combination of loss of situation awareness and skill degradation can result in the operator's inability to respond adequately to the failure of the automation. In the case of the surface movement advisor, these risks are introduced at multiple points in the team structure that includes controllers, pilots, airport managers, and airline operations personnel. On that account, each new automation feature should be evaluated for its impact on situation awareness, all

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation team members should be trained to maintain proficiency in automated tasks whenever they are expected to be able to perform those tasks in response to automation failures, and the capability of team members to manage the complexities permitted by automation should be evaluated. Teams Airport area automation holds the potential for changing the roles of controllers vis-à-vis pilots, airport personnel, and airline personnel. The data distribution and analysis capabilities of the surface movement advisor introduce the potential for realigning responsibility and authority for performing and managing airport operations among controllers, airport managers, traffic managers, and airline dispatchers and analysts. Realignment may include new responsibilities, new authority structures, new communication and cooperative work links, and new measures of effectiveness (e.g., increasing emphasis on efficiency). The impact of the surface movement advisor on individual roles should be considered during advisor analysis, design, and test activities. The teamwork associated with it should also be considered. One promising avenue that can contribute to the design of an effective surface movement advisor is the study of computer-supported cooperative work (discussed in Chapter 3). Such a study should include attention to workload implications of the work requirements and distribution, as well as of the automation of tasks and functions. Effects of Combining Systems As noted in Chapter 5 (with respect to automated ground control systems) and Chapter 8 (with respect to general principles of system development), it is critically important to consider the human factors implications of both phased and simultaneous implementation of two or more automated functions. The combination of automation features can potentially introduce effects that are not predicted from studies or tests of each feature independently. Airport surface automation includes contemplated introduction of many additions or changes to airport operations support tools and, possibly, associated procedures. For example, tools currently available only to some personnel selectively provide such information as airline schedules, flight plans, gate information, various weather parameters, and runway configuration; the surface movement advisor may combine these tools, or future versions of them, and redistribute the information to additional personnel. In addition to the issue of developing a consistent human-computer interface across the integrated tools, their combination may introduce possibilities for redefining tasks. The redefinition of tasks and potentially more timely and accurate information may introduce possibilities for new procedures. The general guidance presented in Chapter 8 applies here: each change introduced should be studied within the operational context, taking into account all

OCR for page 156
The Future of Air Traffic Control: Human Operators and Automation other changes introduced. Such changes may include, for example, the center TRACON automation system final approach spacing tool and the surface conflict avoidance technologies discussed in the previous chapter. The evolution of changes should be centrally monitored and coordinated by a human factors research and development oversight organization.