Pilotless – or remotely piloted – airliners are inevitable, according to industry consensus. The only argument is about how long it will take to get there, with most experts believing it will take a generation or more before the necessary technical integrity brings societal acceptance.
One of the factors driving the belief that pilotless commercial transport aircraft will happen sooner rather than later is industry concern that recent accident statistics show that high levels of flightdeck automation are de-skilling pilots, rendering them ineffective as a fall-back system when the automatics fail. The two possible reactions to this situation are to train pilots more appropriately – which is expensive – or to computerise cockpit functions even further, which is what is already happening.
Aviation consultant Tony Henley, who works with the Royal Aeronautical Society on these issues, has this to say about the pilot expertise dilemma: “There is an unstated question as to whether there is more value in training the crew to be able to deal with failures of the automation or in preventing those failures. Both have significant cost and it might be argued that because, in the event of a failure that is statistically very unlikely, even if the crew has had some relevant training, they still may not respond correctly.
Henley continues: “It might therefore be a better option to make the event less likely even at the risk of making the systems even more complex, which would make it even more difficult for the crew to intervene correctly. Systems don’t make mistakes, but, as yet, are difficult to programme to deal with the unforeseen.”
Independently of this debate, the default industry position on pilots today is to prepare to reduce airline crews to one pilot in the medium term, and eliminate the need for onboard crews in the long term. This position is strengthened by air traffic management development. ATM planners predict a future where, in dense traffic areas, pilots will not have the option of intervening manually because to do so would disrupt the automated traffic separation system. If they may not intervene, why put them on board?
Meanwhile the EU is sponsoring a multinational research programme called ACROSS, which is intended to develop software to enable intelligent automation for automotive, aerospace and industrial applications. Henley, who is advising ACROSS on the regulatory aspects of high level automation, says: “It is relatively easy to automate all the routine functions of the pilot, but the key issues are: who is responsible if something goes wrong? And how does the system deal with the unexpected – for example the Hudson River ditching event?”
The Hudson River incident was precipitated by a birdstrike which rendered both engines on the US Airways Airbus A320 unusable. Ditching was the captain’s choice, and it worked, but there were other possible options. What would an automated system have done? Would it have saved the passengers as the pilots did?
The whole issue of automating pilots out of the system has provoked highly animated discussion on Flightglobal’s Learmount Blog, and most of those who oppose it cite the fact that pilots have had a long history of saving aircraft from disaster, and that these countless successes are ignored in favour of the rare cases where pilot intervention has been useless or even positively harmful.
The commonly cited recent events that demonstrate what pilots are there for – apart from Hudson River – include: the Qantas QF32 A380 catastrophic engine explosion; the British Airways Boeing 777 fuel starvation incident on short final at London Heathrow; and the two celebrated events in which both aircraft suffered total loss of power to control surfaces and had to be flown using engine power only – namely the Sioux City United Airlines McDonnell Douglas DC-10 that suffered catastrophic No 2 engine failure, and the Baghdad DHL A300F that was hit by a missile.
Those are just the spectacular successes, however. The truth about daily line flying is that crews intervene frequently in situations entailing relatively minor faults or risks which, if they had not been dealt with, could have set up a chain of events leading to real danger. But because no-one assembles this enormous quantity of data as a recognised category, the industry has not really considered what level of risk would be created by not having pilots there to deal with apparently minor day-to-day technical or operational incidents.
For the time being, however, ACROSS is looking at the automation systems necessary to support reduced crew workload, reduced crew numbers, and systems for replacing pilot functions in the event of crew incapacitation. The prospect of single-pilot operation for large commercial transport aircraft looms.
Meanwhile a huge amount of work is underway developing non-military partially and fully autonomous unmanned air vehicles, even though ICAO insists that with today’s technology, a remote pilot must be able to intervene or they would not be allowed to fly in common use airspace. On the other hand, one of the most common themes raised by respondents to the Learmount Blog on this subject is the high attrition rate of UAVs, both autonomous and remotely piloted. To say that attrition rates like 30% are unacceptable in commercial air transport is the understatement of the millennium.
The capabilities of future air traffic management systems will play a crucial part in making unmanned aircraft traffic separation possible. Today, although unmanned air vehicles are in common use by the military, they are not allowed to be operated outside war zones or specially restricted airspace because their controlled separation from other traffic cannot – at present – be assured, and their onboard sense-and-avoid anti-collision safety-net systems are not sufficiently developed to guarantee their own separation in uncontrolled airspace.
But in future, according to the chief executive officer of UK air navigation service provider NATS, Richard Deakin, ATM systems will enable controllers to perform the separation assurance function for UAVs as much as they will for manned aircraft. But, he points out, there must be a totally secure communication link between the controller and the UAV’s remote pilot, and that is extremely difficult to guarantee, as is the link between the remote pilot and the UAV.
EASA rulemaking officer Filippo Tomasello agrees with Deakin that providing en-route separation for unmanned aircraft does not require sophisticated equipment, and comments that remote pilots have access to all the sensors that onboard aviators have to detect and avoid other traffic, with the exception of the view through the windscreen. Referring to the datalink through which the remote pilot controls the aircraft, Tomasello points out that it is the weakest and most vulnerable part of the system.
One of the visions for air traffic control in the future is that traffic separation will not be determined on the ground in the same way as it is now – or at least not determined only on the ground. By about 2030, aircraft – piloted or pilotless – will be able to carry out their own separation, using four-dimensional predictive traffic information (the fourth dimension is time) datalinked from the automated ATM system on the ground. This, plus an advanced onboard version of the traffic alert and collision avoidance system, will enable them to thread their way safely through busy traffic, and at the same time to use the most economic route to their destination.
When this level of capability has been implemented, no pilot would be allowed – under normal circumstances – to fly the aircraft manually. Integrated aircraft trajectory management signals and traffic information from the ATM system on the ground would be datalinked to the aircraft flight management system, which would control the autopilot and autothrottle, directing the aircraft. If there were pilots on board, they would be no more than observers to this automatically choreographed progress through the en-route traffic.
Deakin’s general reaction to the proposal for unmanned airliners is to question how the system would cope in the event of an unforeseen disruptive occurrence. He uses the recent real example of the sudden closure of London Heathrow airport when a parked Boeing 787 had an onboard fire. The fire and rescue services were almost fully occupied with controlling the fire, robbing the airport of its required ability to react to an accident, so the airport closed. If, under circumstances like this, unmanned aircraft were already on approach to land, the total system as well as the individual aircraft would have to be capable of adjusting to the requirement for multiple go-arounds followed by diversions.
In a nutshell, Deakin’s hypothetical question highlights the problem that most industry people foresee with the introduction of increasing numbers of unmanned aircraft: will the aircraft, and the systems that support it, be capable of reacting to the unforeseen?
Even foreseeable disruptive events can make life very difficult if they occur suddenly, says Deakin, providing the not uncommon example of a storm cloud drifting across the final approach path at a major airport. Aircraft would have to abandon their approaches and execute a “plan B”. Appropriately trained human beings have the advantage of adaptability in situations like this, despite their human limitations.
André Clothilde, director of the European Unmanned Systems Centre (EuroUSC), has looked into this issue for the Royal Aeronautical Society, and he points out that the questions surrounding it are not new: “In the late '80s/early '90s within NATS we looked at the de-skilling of air traffic controllers and automation of the pilot function to the point where the airborne and ground-based systems were connected and the ‘man-in-the-loop’ reduced to a monitoring and emergency handling function. The ATC controller – not the airborne pilot – would in effect be directing the aircraft, because the ATC had the better situational awareness in a high air traffic density environment and all the flow control information to effect safe navigation of multiple flights.”
Clothilde observes that the key “slow-down” factors to achieving this objective were not technological but “the public, regulators and the acceptance of those already in the loop [ATC officers and pilots]”. He adds: “This tended to make us focus on standards and safety assurance in all its guises being robust enough to meet any agreed reliability figures commonly used for manned aircraft, i.e. the concept of equivalence flavoured with the concept of transparency for existing air operations.”
Now, in 2013, do ATM policy-makers see a future involving the common use of pilotless airliners? Eurocontrol’s Principal Director for ATM Bo Redeborn says: “There is no doubt that it is technically possible at some stage. It could even have some advantages.” He thinks 2030 is unrealistically early, because existing plans for ATM advances in Europe are not being executed at the hoped-for rate, and he does not see this accelerating.
Redeborn says the ATM plans for the future make remotely-piloted – rather than completely autonomous – aircraft “both absolutely possible and feasible”. The two essential ATM capabilities are datalinking systems of sufficient integrity, and the maturity of centralised 4D trajectory management and control. “The system,” he remarks, “is naturally moving in that direction.” Eurocontrol is already preparing for the anticipated arrival of single-pilot operation for large transport aircraft, Redeborn reveals.
He predicts the first mandated use of self-separation will be in oceanic and desert areas, where a combination of ACAS-X (the next-generation of airborne collision avoidance systems) combined with universal automatic dependent surveillance-B extended squitter would enable an accurate real-time and predictive picture of traffic separation to be presented to the aircraft’s FMS and therefore to onboard or remote pilots. This would assure both strategic and tactical separation, and successful performance in these low-traffic sectors would provide the first in-service validation of a system that would eventually be transferred into busier sectors.
The gradual increase in ATM systems capability already means the amount of intervention by ATCOs in the trajectory of flights is decreasing steadily, Redeborn observes, and that process will continue until there is virtually no human intervention at all. When that point is reached, pilots in aeroplanes equipped to be compatible with the new environment would have virtually no options for intervening manually, so there is little point in the pilots being present on the aircraft. Redeborn comments: “It’s bound to happen. But the transition period will take a long time.”
EASA’s Tomasello observes that, as a regulator, it is his job to draw up equipment performance specifications and the legal divisions of responsibility if it fails, not to design the equipment. With the advancement of automation, he sees some legal responsibility shifting from the operators to the equipment manufacturers and software developers, but says the responsibilities of a remote pilot acting as aircraft commander will be identical to that of an onboard captain. Wherever the pilot is located relative to the aircraft, the responsibility for the safety of the aircraft is identical, demanding that the pilot use all means at his/her disposal to ensure the safe conduct of the flight.
Tomasello points out that ICAO is already considering the implications of the wide use of unmanned aircraft, but not yet of totally autonomous flying machines. All its thinking at present assumes a remote pilot will be a part of the system.
He also points out that remote piloting does not necessarily mean one pilot per aircraft or even per flight. Tomasello suggests the possibility of a remote pilot conducting the take-off and early climb at a location from which he/she can see the aircraft, then handing over to an en-route pilot or crew, who themselves hand over again to a remote pilot at the aircraft’s destination. He also proposes a crewing system that would alternately involve an onboard and remote crew for very long haul flights. This hybrid idea could do away with the need to augment the onboard crew for long-haul, because a remote crew could take over during the cruise while the onboard crew rests.
Onboard cabin crew would be less easy to dispose of than onboard pilots, according to Tomasello. He cannot see an emergency evacuation being as well handled by automated systems as by a cabin crew, particularly in their role of preventing panic and promoting orderly behaviour.
Taking a high altitude view of the whole proposal, Tomasello sees movement toward unmanned commercial aircraft as being “extremely gradual”, depending even more on societal acceptance than technological advances. “It may happen in 50 years,” he says, “but I wouldn’t bet on it.”