Automation in Aircraft: The Changing Role of the Pilot
The aviation industry has always been at the forefront of engineering and technological innovation. It has developed alongside the computer revolution in the 20th century and aircraft manufacturers have been quick to take advantage of the increased reliability and accuracy of automated systems compared to those operated by humans. Aviation has also benefitted from the research and development budgets afforded to military and space programmes, driven by two World Wars and the subsequent Cold War. Airframers and engine manufacturers to this day often work on civilian airliners alongside military projects, with crossover of technology between the two. In the last three decades automation in aeroplane cockpits has increased hugely whilst training of pilots has in general retained a more traditional emphasis on stick-and-rudder skills. Some recent incidents have highlighted that increased cockpit automation without appropriate training may in fact be increasing the risk of accidents, and airlines are starting to incorporate this factor into training programmes.
From the Wright Flyer to the modern jetliner
When aeroplanes were first conceived early pioneers such as the Wright Brothers relied entirely on mechanical connections between cockpit controls and the moving parts of the aircraft, with a system of cables and pulleys connecting the yoke or stick and rudder pedals to the three primary control surfaces of the machine: ailerons, elevator and rudder. Basic piston engines with fixed-pitch propellers as fitted to early aircraft as well as many light aircraft built and flown today require only mechanical throttle, mixture and carburettor heat controls. Early aeroplanes could only fly low and slow, in visual conditions, so the workload on pilots was sufficiently low that they were able to manage the controls single handedly. The main concern would have been mechanical failure of the structure or the engine, sadly commonplace in the first years of aviation.
As aircraft complexity grew and reliable computer systems became available, manufacturers incorporated increasing levels of automation into their designs. Automation in modern aircraft can be found virtually everywhere, from FADEC (Full Authority Digital Engine Control) systems controlling the engines, through fly-by-wire electronic movement of the aeroplane's control surfaces, to navigation and autopilot systems accurate enough to land over a hundred tons of metal flying at over 150 miles per hour on a narrow strip of tarmac with no pilot intervention other than to lower flaps and undercarriage. The role of the pilot clearly must be different in these ultra-modern aircraft, and indeed if it wasn't for those pesky flaps and gear the aeroplane could fly itself with no pilot at all. Military UAVs do just this; controlled by a pilot on the ground, all on-board systems are automatic. Most people feel happier having pilots in the pointy end, though, and there are currently no serious plans to remove humans from the cockpit of commercial airliners.
Boeing 747 over the years
The Boeing 747 was conceived in the early 1960s initially to serve the US military's cargo transport requirements. However, commercial jet travel quite literally took off around this time, and the company saw a niche in the market for a large passenger aircraft. The first variant, the 747-100, made its first flight in 1969 and entered service with Pan Am in 1970. As can be seen in the top image on the right, the flight deck was full of traditional analogue-type instruments and gauges, and flying the aircraft required three crew members - the captain, the first officer and a flight engineer, who sat behind the pilots facing a huge instrument panel monitoring and controlling many aircraft systems.
The flight engineer was rendered obsolete with the advent of the 747-400 variant in 1989. The second picture shows the completely different layout of the cockpit, with large digital electronic displays in front of the pilots. Many functions that had to be performed manually were incorporated into automatic sequences activated by the push of a single button, reducing the number of dials, gauges and knobs in the cockpit from nearly 1000 to just 365.
In 2011 the latest 747-8 variant entered service with updated systems using technology developed for Boeing's all-new 787 airliner. The cockpit shown in the third photograph looks similar in terms of displays, hiding improved systems and use of fly-by-wire for some control surfaces.
Fly-by-wire is a system that replaces mechanical connections between the pilot's controls and the aircraft's control surfaces by an electronic interface which interprets the pilot's inputs and converts them to electronic signals which cause actuators to move the surfaces appropriately. Such systems can be used to improve flight stability, correcting automatically for disturbances from the desired flight state. This is used particularly in modern fighter jets which are highly manoeuvrable but inherently unstable and would be unflyable by a human if the flight control computers failed. Airliners do not behave in this way, but flight control computers can be programmed for flight envelope protection. This means that if a pilot's input demands a manoeuvre that would put the aircraft in a dangerous position or exceed structural limits, the computer will only allow the manoeuvre up to predetermined limits, whilst giving the pilot a cockpit warning.
Fly-by-wire has an impressive history, with Concorde the first production airliner to make use of it in its early analogue form, and the Space Shuttle the first aircraft to use a fully-digital system. Airbus was the first major civil airframer to adopt digital fly-by-wire technology, using it to a limited degree on the A310 and fully on the A320, introduced in 1988 and known colloquially as the "Electric Jet". After some initial resistance Boeing has subsequently adopted the technology and currently uses it on the 777, 787 and 747-8. Embraer uses fly-by-wire for its E-Jet series, and the Dassault Falcon 7X is the first business jet to use the system.
Air France Flight 447
On June 1 2009 Air France Flight 447 from Rio de Janeiro to Paris crashed into the southern Atlantic Ocean. Initially the cause of the accident was a mystery as there had been no mayday call from the pilots and recovery of the flight data recorders from the seabed took two years. Investigations focussed on suspected icing of instruments called pitot probes due to the weather conditions at the time. Normally pitot probes are heated to prevent ice formation, but if they become blocked the aeroplane's computers cannot accurately tell how fast it is travelling through the air. There is then a risk of flying too slowly, meaning there will not be enough air passing over the wings to maintain lift, and it will enter a stall and fall out of the sky. Discrepancies between readings from different pitot probes makes it impossible to know which is the correct airspeed, and the autopilot which would have been flying the Airbus A330 at the time responded by disconnecting, meaning that the pilots were flying by hand. Critically, in this situation the flight control computers are programmed to degrade from "normal law" to "alternate law", in which condition flight envelope protection is not available and there is nothing to stop the aeroplane entering a stall, which is what happened in this case.
The mystery was why the crew were unable to recover from the stall. The standard stall recovery procedure for any aeroplane is to lower the nose to increase speed and allow the wings to generate lift again. Flying at over 35,000 feet there should have been plenty of time to recover, but it did not happen.
Recovery of the flight data and cockpit voice recorders revealed what happened, in chilling detail. It was clear from the pilots' dialogue that they had no idea what was happening until it was too late. Worse, one of the pilots was responding entirely inappropriately to the stalled condition by pulling back on the stick rather than pushing forward, despite stall warning indications in the cockpit. As with almost all aviation accidents the cause was multifactorial with contributions from the weather, the design of the pitot probes, and inappropriate pilot response to loss of automation.
Colgan Air Flight 3407
A few months earlier, on 12 February, a Bombardier Q400 crashed on approach to Buffalo, New York. The investigation revealed this aircraft had also stalled, albeit under different conditions. The autopilot was flying the aeroplane but the airspeed was falling rapidly as flaps and undercarriage were lowered, increasing drag. The crew failed to notice this until the stall warning system activated the stick-shaker. The appropriate response would have been to carry out a standard stall recovery, lowering the nose and applying power to increase speed. Instead the captain tried to pull the nose up. This activated the next stage of stall protection, the stick-pusher, which applies a strong nose-down force to the stick to recover the aircraft. However, the captain continued to pull back against this force, resulting in a full stall from which the aeroplane did not recover.
Accident prevention through pilot training for the modern cockpit
In both these cases deficits in training emerged as the investigations progressed. As a result of the Colgan crash it emerged that stall recovery as generally taught in the simulator emphasised recovery with minimum height loss, meaning that pilots were tending to move away from the stall recovery technique that they would have learnt early in their training in light aircraft, with a definite nose-down movement, and were tending to avoid height loss at all costs. This was thought to have contributed to the captain's inappropriate response to the stall warnings.
The Air France investigation discovered that the airline did not include any high-altitude stall recovery procedures in its training programme, and indeed no manual handling high-altitude training at all. This is likely to have resulted in the confusion shown by the pilots, their failure to follow appropriate checklists and their inappropriate responses.
These accidents have provoked much discussion in the aviation industry. In 2011 a report was presented at the Royal Aeronautical Society flight training conference suggesting that flight crew able to respond to unfamiliar situations often involving loss of automation were either military-trained or employed by an airline with more than the legal minimum recurrent training programme, suggesting that standard training is not providing modern pilots with the resilience required to manage modern cockpit challenges. It has been suggested that newly-trained pilots are relying too much on automation, unlike older pilots trained on less sophisticated systems who tend to question sources of information and be better prepared for malfunctions of any system.
There have been some immediate changes in training, notably in stall recovery where there has been a return to basic principles - nose down and increase airspeed as a priority. Some flight training organisations have introduced an upset recovery module into their basic training to increase situational awareness and instil correct recovery procedures at an early stage. There is a move towards competency-based training through the introduction of a new licence, the Multi-Crew Pilot Licence, which is designed to assess trainees based on actual performance in a range of situations rather than simply completing an arguably outdated syllabus.
It is unfortunate that sometimes accidents have to happen before problems are identified, but awareness of the different challenges to pilots operating highly automated aircraft is now at an all-time high, hopefully resulting in increased safety for all who fly.