The User Interface of Britain's New En-Route Centre for Air Traffic Control
Jim Cozens
Catcon Limited,
The Chart House, Ballfield Road,
Godalming, Surrey,
GU7 2HA, UK
INTRODUCTION
For the past 6 years National Air Traffic Services and their prime contractor Lockheed Martin have been engaged in the development of a new Air Traffic Control Centre that will control aircraft 'en-route' over England and Wales - essentially planes flying above15,000 feet. The new centre, known as 'NERC', will replace the current centre at West Drayton and is expected to increase air traffic capacity by 40%. This paper discusses four key decisions in the user interface design. For each we describe the design problem and the rationale for the chosen solution and give an assessment of how the system is working in practice.
DISPLAY MANAGEMENT
The 'on screen' part of the NERC user interface is a GUI controlled with mouse and keyboard. One of the first issues in it's design was how to organise the displays. To be successful in a real time environment we needed rapid access with the minimum overhead in managing windows. The familiar 'desktop' solution with overlapping windows has too much 'swapping' for this contextso we chose an approach we called the 'workbench' in which primary data is always present, with 'tools' and additional data quickly accessible. The primary data is provided by 'permanent windows' - the radar plus a border on the main display, an airspace map plus border on the auxiliary - that tile the display, overlaid by movable windows containing flight data in 'electronic strip bays'. Other windows are accessed via buttons in the borders and/or directly through the primary data. This design does have one awkward compromise. If the sector for which a controller is responsible is wide but short, having the tall, thin strip bays on the main display means that the radar is used at a smaller scale than is desirable. The preferred alternative is to move the strip bays to the auxiliary display, reducing the effective size of the airspace map and so slowing down access to support information. In the usability trials and others large system tests we have run, the display organisation attracts little comment except for the above issue (no comment is a complement from our vociferous users). Our observations show that the controllers spend little time managing their displays once they have them set up (some button labels need improving) so we think this aspect of the user interface is a success.
INTERACTION STYLE
In the current en-route centre, the controllers use paper flight progress strips that they update by hand. Although apparently crude, this is a highly efficient system. Early in the project NATS decided that the 'tactical' controllers (who talk to aircraft) would continue to use paper strips for their moment by moment record. The NERC computer systems would, however, hold the 'coordination' data (the plan for where aircraft enter and leave individual sectors of the airspace)which would be entered by the 'planner' controllers and used by both tacticals and planners. To be effective, the NERC user interface has to be as efficient as pen and paper for entering coordinations. The interaction style we chose is a combination of mouse plus keypad (number pad plus function keys). We chose the mouse for its all round strength as a pointing device, the number pad because the main data to be entered is 3 digit levels (the menu
selection alternatives we looked at were typically less efficient and/or more error prone) and function keys for commands that had no simple direct manipulation equivalent. We are currently seeing two problems with the interaction style: the controllers are taking time to become assured with it and for a few key tasks the interactions are clumsy. We are confident that the first of these is simply a matter of practice - the users with the most experience are now using the system in an assured and skilled manner. The unwieldy interactions are being refined and we expect to improve themto the usability level we already have for the majority of key tasks. INFORMATION MANAGEMENT
A primary ATC technique is the 'scan' in which the controller regularly examines the data for the aircraft for which he/she is responsible to maintain a mental picture of the traffic situation. This approach uses information 'remembered in the world' and demands that all the data being used is displayed all the time - interacting with the system during the scan is unduly disruptive. Moreover, what is needed is not only the task data but also contextual data to help maintain the structure of the model. Key features of the user interface design to support this method are: different data displays provide the pertinent data for each members of a control team - the data needed by the tactical controller is on the radar plus the paper strips, the planner's data is on the electronic strips plus the radar, and specialised arrival and departure lists allow assistants to communicate with airfields; the individual displays are configurable - a variety of data can be selected for display in radar track data blocks, the electronic strips have collapsed and full forms and can be sorted in half a dozen different ways; and additional data is available on demand - global 'quicklooks' provide one extra datum on every radar track datablock (which fits the scan) and a mouse driven 'pop- up' provides full data on one track or flight plan. Usability trials have not uncovered any substantial problems with this approach but we have identified further applications of 'quicklook'.
ALERTING AND ATTENTION GETTING
Modern air traffic control systems aid the controller by detecting potential problems before they occur. Both these alerts and unsolicited changes to the data on display need the controller's attention but may not be as urgent as his/her active task. To balance the need to alert with the need to avoid interruptionthe NERC user interface uses different forms of display according to the importance of the data: simple status indicators are black and steady, eg the buttons in the border that bring the strip bays onto the display show an icon indicating whether there are strips in the corresponding bay; changes to data on the electronic strips 'throb' between black and dark grey - this gives a low priority 'notice when you look closely' alert; alerts that the controller is likely to attend to as the next task use colour infills or hatching, eg if a planning action is overdue, the corresponding strip is marked in orange - these indications are immediately noticeable when the controller scans the display, may distract'; flashing is reserved for the one truly urgent system event - the short term conflict alert. The design goes outside this scheme for parts of the workstation that are outside the main visual field: eg flashing is used on the telephone panel for incoming calls and to indicate outstanding messages in the system message area in the top border of the main display. Our preliminary assessment is that scheme this works as intended. However, this facet of the design is susceptible to the affects of high workload so we will be monitoring itas we run high workload tests in the run up to operation.
