Flight Deck Automation

Flight deck automation is the set of systems that carry out tasks the pilot would otherwise do by hand: the autopilot and flight director, the autothrottle or autothrust, the flight management system (FMS), and the envelope protections of fly-by-wire aircraft. It flies more precisely than a person can, spreads the workload, and frees capacity for planning, monitoring and communication.

It also changes the pilot's job. Automation does not remove workload so much as change its nature, replacing active control with passive monitoring, which is the task humans perform worst. Much of what goes wrong on modern flight decks happens at this boundary: a mode the crew did not expect, a speed nobody was watching, a disconnection that hands a confusing situation back to a pilot who has not flown by hand for weeks. Automation is part of CRM training under EASA's ORO.FC rules, and automation management is one of the components of the FAA's single-pilot resource management.

The role of automation

About 70 % of aviation accidents are attributed to human error, and that figure has driven the spread of automation, as the industry looks for ways to remove the human element where it is safe to do so. The benefits are real. An autopilot engaged early in a busy arrival releases mental capacity for briefing, checklists and threat assessment, and an autopilot holds a flight path more precisely than hand flying, especially in IMC.

Human performance texts list the disadvantages that come with it:

• Delay between action and effect. An incorrect waypoint entered into the FMS may not show itself until hours later.
• Crew coordination. The pilot monitoring moves less, so there are fewer body-language cues; information is abbreviated into acronyms; and each pilot has independent access to data and commands without necessarily telling the other.
• Boredom. Long automated sectors bring loss of situational awareness, hypovigilance and loss of handling skills.
• Blinkered concentration. A pilot can become so absorbed in one readout that situational awareness is lost, when a cross-check against another instrument would resolve the doubt.
• Failures. They are less frequent than before, but less predictable, much quicker and more intense when they come.

Levels of automation

Automation is not simply on or off. It is used at different levels:

On the Airbus flight control unit, pulling a knob engages selected guidance with the value in the window as the target, and pushing it engages or arms managed guidance from the flight plan. Boeing's mode control panel offers equivalent choices between modes such as heading select and LNAV.

No level is best in itself. Airbus puts the principle into its golden rules: the crew must use the appropriate level of automation at all times, which can include manual flight, and must understand the operational effect of the level selected. A decision to fly manually is agreed between both pilots and based on the aircraft's status, pilot fatigue, the weather, the traffic and familiarity with the area. When the aircraft does not follow the desired path, the pilot flying changes level, from managed to selected guidance or from selected guidance to manual flying, while the pilot monitoring communicates, challenges and takes over if necessary.

Managing automation

The flight mode annunciator (FMA) is the primary place where the aircraft shows which modes are actually engaged and armed for thrust, pitch and roll. A knob push or pull is a request; the FMA is the answer. Airbus's rule is to monitor, announce, confirm and understand the FMA at all times, and SOPs have every mode change called out and checked by both pilots against what was intended.

View full schematic

Good automation management rests on a few habits:

• Verbalise and verify. Announce each selection, confirm the resulting mode on the FMA, and state what the aircraft should do next.
• Cross-check raw data. Compare the FMS picture with raw navigation data, especially on approach.
• Share the head-down work. Programming the FMS in a busy phase takes one pilot's eyes off the flight path. At Kegworth in 1989 the first officer spent two minutes of a high-workload descent trying without success to reprogram the FMS for the diversion airfield.
• Step down when surprised. When a crew finds itself asking what the aircraft is doing now, the answer is to reduce the level of automation, disconnecting the autopilot or autothrottle if necessary, fly the aircraft, and re-engage only once the logic is understood. Many loss of control events began with a crew reprogramming an FMS instead of flying.

Automation complacency and dependency

Automation complacency is over-reliance on automation. Its classic symptom is passive monitoring: leaving it to the computer, in the belief that the automation is more capable, until the pilot regards the aircraft as infallible. Sustained monitoring of a system that is usually right is a task humans do badly. Trust grows, the sampling of mode annunciations slows, and the pilot ends up discovering a problem from the flight path instead of from the annunciator.

Automation dependency is reliance on automation to the point where the pilot can no longer fly confidently without it. The FAA's SAFO 13002 notes that continuous use of autoflight can degrade a pilot's ability to recover quickly from an undesired aircraft state. The NTSB found that Asiana's automation policy emphasised full use of automation and did not encourage manual flying, and that more manual practice would likely have helped the pilot flying notice the decaying speed and add power.

Mode confusion and automation surprise

A mode error, or mode confusion, occurs when the pilot believes the automation is in one mode while it is actually in another. An automation surprise follows when the aircraft then does something the crew did not expect. Keeping the modes in the central scan is the defence.

Asiana Airlines flight 214 is the standard case. On 6 July 2013 a Boeing 777 was on a visual approach to San Francisco, high on the profile. The pilot flying selected flight level change mode while below the selected altitude, so the autopilot began to climb and the autothrottle advanced the thrust. He disconnected the autopilot and pulled the thrust levers to idle, and the autothrottle changed to HOLD, a mode in which it does not control speed. None of the three pilots later recalled seeing the change on the FMA. The NTSB found the autothrottle logic complex and not intuitive; interviews showed that other pilots at the airline, including some instructors, held similarly inaccurate mental models of it. It found the crew's monitoring of airspeed insufficient because of expectancy, workload, fatigue and reliance on automation. The aircraft struck the seawall; three passengers died (see stabilised approach).

On Turkish Airlines TK1951 at Amsterdam in 2009, a faulty left radio altimeter reading of -8 ft put the Boeing 737's autothrottle into its RETARD flare mode, closing the thrust levers on the approach. The autopilot kept following the glideslope by raising the nose as the speed decayed, and the crew did not notice the lack of thrust until the aircraft neared the stall; nine of the 135 occupants died (see radio altimeter). The lesson common to both: read the FMA after every change, and keep speed and thrust in the scan on every coupled approach.

The out-of-the-loop problem

A pilot who is only monitoring is out of the loop: no longer controlling the aircraft, less aware of its state, and often not mentally prepared to take over when the automation fails or disconnects. The handover then comes at the worst moment, with an abnormal situation and an alarm. Startle is the involuntary reflex to a sudden intense stimulus; surprise is the slower cognitive response to an event that violates expectation, and the mental effort of rebuilding a picture that has just been shown to be wrong. The first response to either should be to stabilise the flight path and buy time, not to act on the first explanation that comes to mind.

Air France flight 447 showed the cost. On 1 June 2009 an Airbus A330 cruising at FL350 lost consistent airspeed indications, most likely because ice crystals obstructed the pitot probes. The autopilot and autothrust disconnected and the flight controls reverted to alternate law. The pilot flying's nose-up inputs took the aircraft to about 38,000 ft and into a stall that the crew never identified, and all 228 people on board died. The BEA listed among the factors the absence of training at high altitude in manual handling and in the unreliable airspeed procedure, and task-sharing weakened by incomprehension of the situation and poor management of the startle effect (see unreliable airspeed).

Maintaining hand-flying skills

Manual flying skills decay when they are not used, and they are needed most when the automation has just failed, disconnected or been set up wrongly, often while the pilot is also surprised and troubleshooting. The FAA's SAFO 13002 and SAFO 17007 encourage operators to give pilots opportunities for manual flight when conditions permit, and a balanced practice of hand-flying departures and approaches keeps the skill available.

The timing matters. A clear day, light traffic, a familiar airfield and a rested crew make a good opportunity; a complex arrival with frequent ATC changes does not. After AF447 the BEA recommended regular training in manual handling of the approach to stall and stall recovery, including at high altitude, and in handling surprise. Regulators have since made upset prevention and recovery training mandatory for airline pilots.

Hand-flying also counters underload. On a long, quiet automated sector arousal falls, attention wanders and vigilance for rare events decays; periods of manual flying, alongside a disciplined scan and regular systems checks, bring arousal back towards the middle of the performance curve. For the single pilot flying IFR, the autopilot remains a workload tool rather than a substitute pilot: it should be engaged early in a busy phase while the pilot stays actively in the loop, and it never replaces instrument currency.

Library articles are written for study and exam preparation. They do not replace your aircraft's approved documentation, your operator's procedures or the regulations themselves.