David Learmount/LONDON
TODAY'S "GLASS COCKPITS" are designed, using a flawed concept and are causing pilots to make mistakes, which they have never made before, according to recent research. Yet flight decks of basically the same design will be built for many thousands more commercial air-transport aircraft during the next 15 years or so.
These highly automated flight decks, fitted with sophisticated flight-management systems (FMS), have caused an "inadvertent change in the role of the pilot", say researchers at NASA's Ames and Langley Research Centers in the USA. The role change makes the pilot more of a monitor and less of a commander.
So serious is the issue that Victor Riley of the Honeywell Technology Center calls for "revolutionary, rather than evolutionary" improvement. This does not seem likely to happen in the near future.
The concerns are not new. In October 1994, recognising the need for a fundamental review of flight deck safety in "highly automated" airline cockpits since their introduction in Boeing's 757 and 767 and Airbus Industrie's A310 in the early 1980s, the US Federal Aviation Administration set up the Human Factors Working Group (HFWG). This FAA-led international commission has now completed its research and is preparing a report, which is expected in late May. Meanwhile, in February this year, the UK-based Royal Aeronautical Society (RAeS) held a seminar in London, entitled: The future flight deck: safe and user-friendly?
In practice, the seminar concentrated more on defining what has been done badly in today's highly automated cockpits than in what the future holds.
Riley's call for a revolution in pilot/aircraft-interface design contrasts strongly with actual industry practice. The flight deck design philosophy exemplified in the industry's latest product, the Boeing 777, is intentionally evolutionary. Although the aircraft is Boeing's first fly-by-wire aircraft, the company's 777 cockpit design philosophy was to present pilots with an environment which did not contrast too strongly with that to which they were accustomed in conventionally controlled aircraft. Another aim was to minimise type-conversion training from other glass-cockpit Boeings - and even from traditional types. In the 777, Boeing uses avionics suites and interfaces which, it may be argued, incorporate some advances, but also most of the disadvantages which NASA and others criticise. Riley, however, is looking at the future. In the 777, Boeing was contending with the present.
Even Airbus, except for its introduction in 1987 of the side-stick in the A320, cannot be said to have been revolutionary in the A320's general cockpit design. Use of the cathode-ray-tube (CRT) as the instrument for presenting flight and systems information had already been established. Airbus merely used the CRT, plus digital-processing technology, as a means of taking the display-integration process a step further and making more information available. At about the same time, similar avionics suites for other glass cockpits were being prepared for the Boeing 747-400 and the McDonnell Douglas MD-11.
ACCLAIMED CHANGES
Airbus introduced two universally acclaimed changes, however, which will probably - in concept at least - not be altered, if Riley's pleas for more "intuitive" information on the flight deck are ever heeded. One is the speed-trend arrow on the airspeed indicator (ASI). The other is the presentation in graphic/pictorial form of systems information on the electronic centralised aircraft-monitor (ECAM). Airbus' introduction of the prioritisation of ECAM display and alarms by phase of flight was also arguably revolutionary.
A significant number of pilots argue, however, that the speed-trend indicator was a step forward to compensate for a step back which all the manufacturers had taken. This "retrograde" step was the introduction of the "strip" or "tape" display to replace the dial-and-needle displays for ASIs and altimeters. The argument is that, unlike strip displays, the dial displays are graphically unambiguous, and needle-movement rate provides the speed-trend information with no need for enhancement.
The issue is clarified by a particular example of the dilemmas faced by the cockpit designer: Airbus senior vice-president of engineering, Bernard Ziegler, pointed out to the RAeS that, in developing the ASI's tape display, there were ergonomic arguments of apparently equal validity as to which way the tape should "move" relative to the ASI pointer, which is static on the primary flight display. Should the tape be seen to move upward as speed increases? Airbus - and all the others - decided that the tape should move downward, so that the pointer should appear to be moving up the tape, and the numbers visible on the tape as it scrolls up or down would be higher at the top of the ASI display. The same argument was adopted for the altimeter strip, which, of course, means that the tape is seen to move downwards as the aircraft climbs, and upwards as the aircraft descends.
WHAT "INTUITIVE" MEANS
It can be argued that the movement of a needle clockwise or anti-clockwise is not intuitive. Honeywell's Riley, however, gives an explanation of "intuitive" as being something which seems natural by virtue of what people have learned in everyday life. He cites desktop-computer manufacturer Apple Computer's symbol on its Macintosh for the deletion or dumping of a file as being intuitive: an Apple Mac screen shows a symbol of a waste bin, which is placed on the "floor" of the display: a file is dumped by identifying it and moving it into the bin. Waste bins are intuitive symbols to those who have used real ones.
The "inadvertent change in the pilot's role" to which NASA refers has been a secondary effect of the extensive use of digital technology for cockpit automation. When it was introduced, the automation was intended to reduce pilot workload and allow more accurate flight, but also - vitally important at the time - to enable wide-bodied aircraft to be operated without a flight engineer for the first time. Although secondary-effect changes can be benign, this "inadvertent change" was not a happy one, according to NASA Langley-based Dr Kathy Abbott, who explains: "The current [flight deck] design approach [has created] the potential for new types of human error." FMS mode-confusion, or lack of mode-awareness, is foremost among the "new errors" named by Abbott. She adds, however, that pilot tasks are now often disjointed and difficult to integrate, and that "subsystems management", like the manipulation and updating of the control/display unit, add to the pilot's tasks.
Aerospatiale's Jean-Pierre Daniel, who worked on the A310's cockpit design some 15 years ago, recalls overpowering pressure from the airlines to automate the flight deck to the point where there could be no plausible aircrew-union argument for retaining a flight engineer, even in complex wide-bodied aircraft. Boeing, meanwhile, had designed a three-crew 767 flight deck, as well as the two-pilot one, in case unions or regulators forced adoption of the former. This pressure was by no means the only motivation for the introduction of digital technology, but Daniel nevertheless stresses its significance. In 1994, after more than a decade of first-generation glass-cockpit operational experience, Daniel commented: "Pilots have to be able to act in order to maintain awareness. The question is what should they do and how much? Today the [pilot] workload is certainly not too much. Perhaps we have to increase it."
HUMAN-CENTRED DESIGN
Research at NASA today, says Abbott, is intended to "...re-assess the flight deck design process" on the premise that "...a human-centred systems-oriented-design approach will reduce human errors and improve safety". The implication is that current flight decks are not designed with a "human-centred" philosophy. In fact, NASA describes them as "technology-centred". A second implication is that a replacement for today's much-criticised cockpit will not be ready for many years, so airline pilots will for a long time have to make the best use of what they have.
UK Civil Aviation Authority test pilot Capt Terry Newman says that getting the optimum performance from a computerised cockpit is a complicated task. He points out that automation technology is complex in all respects: its hardware, software, inputting disciplines, mode differences, glitches and lack of transparency. It confronts the pilots with more choices than they have ever faced, requiring more decisions. This, in turn, has led Newman to state that airline pilots are given inadequate type-conversion training for highly automated aircraft.
Abbott confirms this, saying that increased automation demands "extra time for transition [type-conversion] training" and that there is an increased failure rate in type-conversion training. NASA also observes that pilot effectiveness in the automated cockpit involves "a long learning curve - about a year on-the-job-training". The agency states, that the most common question asked of it by airlines is: "What do you know that will help with training for automation?"
DANGEROUS ASSUMPTIONS
Although Newman says that the views he presents are his own, he has been a European Joint Aviation Authorities representative on the FAA's HFWG. Commenting on training for highly automated aircraft, Newman points out:
the assumption continues to exist that the crew is properly competent, in the face of increased complexity and the need to be familiar with both manual and automatic flight-path management, but the time allocated for training has not kept pace;
either an investment in more training is essential, or more strict rules for system design must be enforced;
automation, particularly in the vertical-flight-path-management modes, is having to be "...manipulated in a manner which has not been envisaged by the designer", because air-traffic control is demanding increased accuracy in four-dimensional navigation [the fourth dimension is time], in the knowledge that adept use of the FMS by the pilots can normally provide it;
there is competition between manufacturers to demonstrate that the ease of operation - by virtue of automation - of an aircraft should enable training time to be "...further reduced".
Newman concludes that training for cockpit automation is needed, to overcome shortcomings which tomorrow's systems should not have, but says that "...[training] is the only method that has enough flexibility in the short term to deal with the design issues that are troubling us".
Honeywell's Riley offers an analogy for the FMS task compared with what he intends that the future should offer. Operating today's FMS, he says, is like operating early personal computers run entirely on raw DOS software: it is not intuitive and demands extensive training. The future, Riley says, should look for a quantum leap to the FMS equivalent of commercial hardware/software packages such as the Apple Macintosh or Microsoft Windows, providing a user-friendly, intuitive interface.
Despite all the criticisms of highly automated aircraft, which includes almost all in-production airliners, the latter have a lower accident rate than those of their forebears, according to Boeing research. Other sources of information pose thought-provoking, but unanswered questions. NASA says that data from its extensively used US Aviation Safety Reporting System (ASRS) indicates that 60% of reported altitude deviations took place in glass-cockpit aeroplanes, the remainder in traditional types. Cockpit type did not affect ASRS-reported track deviations.
Roger Green, chief scientist at the UK Defence Research Agency's Centre for Human Sciences, reports that pilot responses to an extensive questionnaire about the nature of their work showed that reaction varied far more between pilots from different airlines operating identical types than between pilots from the same airline operating different types - including differences between glass-cockpit and traditional-cockpit aircraft. Company approaches to training may be the key.
Airbus' Ziegler points out that engineers can always develop control designs which apparently have an impeccable logic to the point of being apparently foolproof. Meanwhile, pilots, not always given the time to apply cool logic, will still sometimes interpret the selection of a control-lever position as being the execution of the command when, as Ziegler points out, "...it is only confirmation that the command has been given". He gives the example of the moving auto-throttle lever being interpreted as evidence that speed control is being maintained, or flap lever selection to a chosen position as evidence that flap has been deployed. His conclusion is that systematic analysis of real experience with existing systems and their improvements is one of the most important ways forward, because the environment in which pilots have to operate is not like a laboratory.
IMPROVEMENTS POSTPONED
It seems clear that, although the airline world can easily criticise the automated cockpits it has today, there will be no sudden change to a magic "human-centred" design in the next decade, because the very nature of a human-centred design and what it means in practice is still being researched. It will probably involve the pilots more physically, certainly more consistently, in the operation of the aircraft, and will give them more immediate and intuitive feedback than the current systems do.
Meanwhile, industry will have to rely on data from the HFWG report, which, it is hoped, will be published at the end of May, and improved industry information exchange, to determine how to make the best of an imperfect system for many years yet.
Many (but not all) of the statements, findings and data in this article were presented at the Royal Aeronautical Society's February 1996 conference in London, entitled: "The future flight deck: safe and user-friendly?"
Source: Flight International
