What is an aircraft autopilot, and how does it work?
Learn how an airplane autopilot works, what its modes control, how pilots use it, and why it may not follow the route, altitude or speed expected.
An aircraft autopilot is an automatic flight-control system that moves the aircraft's flight controls to follow selected targets such as heading, altitude or a navigation path. It continuously compares sensor data with the commanded target, calculates any error and applies corrections; pilots still choose the modes, manage thrust where required and monitor the result.
Within our Aviation & Real-World Flying coverage, autopilot means the real aircraft system. Flight simulators reproduce the same control principles with varying levels of system depth, sensor modelling and mode accuracy.
How does an airplane autopilot system work?
An airplane autopilot works through continuous feedback loops that compare the commanded flight path with the aircraft's measured movement.
- A target and mode are selected. The pilot might select a heading, altitude, vertical speed or airspeed. A flight management system can instead supply route or vertical-path guidance.
- Sensors measure the aircraft's state. Attitude and heading systems report pitch, bank and direction. Air-data systems supply altitude, airspeed and vertical speed, while navigation receivers provide GPS, VOR, localiser, glideslope or other guidance.
- The flight-guidance computer calculates the error. An outer control loop might convert a route deviation into a required bank angle. An inner loop then calculates the aileron, spoiler or equivalent command needed to achieve and hold that bank.
- Actuators move the flight controls. A conventional installation may use electric or hydraulic servos connected to the control system. In a fly-by-wire aircraft, autopilot commands pass through the electronic flight-control computers.
- Feedback confirms the result. The sensors measure the aircraft again, allowing the computer to reduce, increase or reverse its correction. Automatic trim may remove sustained control loads.
This cycle runs continuously while the autopilot is engaged. If required sensor data become invalid, control forces exceed limits or the system can no longer follow its command safely, it may refuse to engage or disconnect with an aural and visual warning.
An autopilot has finite control authority and operating limits. A basic system should not be assumed to prevent a stall, overspeed, excessive bank or flight into hazardous weather; any protections depend on the particular aircraft and active mode.
What can an aircraft autopilot control?
An aircraft autopilot can control one or more axes, with capability ranging from simple wing levelling to coupled lateral and vertical navigation.
- Roll: ailerons and, on some aircraft, spoilers control bank angle, heading, route tracking and localiser capture.
- Pitch: elevators and stabiliser trim control pitch attitude, altitude, vertical speed, airspeed-based climbs and vertical paths.
- Yaw: the rudder can provide directional control, although many aircraft use a separate yaw damper for turn coordination and Dutch-roll damping.
A single-axis autopilot normally controls roll. A two-axis installation adds pitch, while a three-axis installation also controls yaw. The number of axes is not the number of autopilot channels: multiple channels are independent or redundant control paths and may be required for an automatic landing.
| System | Primary job | Can work independently? |
|---|---|---|
| Autopilot | Moves or commands the flight controls to follow active guidance modes | Its available guidance depends on the installation |
| Flight director | Displays pitch and roll commands for the pilot to follow manually | Yes, it can remain in use with the autopilot off |
| FMS or FMC | Stores route and performance data and generates navigation guidance | Yes; it does not normally move the flight controls itself |
| Autothrottle or autothrust | Adjusts engine thrust to meet speed or power targets | Usually; it can remain engaged when the autopilot is off |
| Yaw damper | Applies small rudder corrections to damp unwanted yaw | Usually, depending on the aircraft |
These systems are closely integrated in an airliner, but they are not interchangeable. Our explanation of why the FMC plans the route while the autopilot controls the flight path covers a distinction that causes frequent confusion.
How do autopilot modes decide what the plane actually does?
The active lateral and vertical modes determine what the plane does; selecting a value or pressing a panel button does not necessarily activate the expected mode.
- Lateral modes include roll or wing-level hold, heading select, navigation tracking and localiser capture. Typical labels include
ROL,HDG,NAVandLOC. - Vertical modes include pitch hold, vertical speed, altitude hold, airspeed-based climb or descent, vertical navigation and approach-path capture. Labels may include
PIT,VS,ALT,IAS,FLC,VNAV,GSorGP.
Mode names and capture logic vary between manufacturers and avionics suites. An armed mode is waiting for its capture conditions; an active mode is controlling the aircraft now. Navigation mode, for example, may remain armed until the aircraft approaches the route at a suitable intercept angle.
The decisive indication is the Flight Mode Annunciator, usually displayed near the top of a glass primary flight display. A mistake we see constantly is watching an illuminated button instead of reading the active and armed mode annunciations.
Selecting a higher altitude while altitude hold remains active will not necessarily make the aircraft climb. The pilot must also select or confirm a suitable vertical mode and manage thrust. As the aircraft approaches the target, altitude capture normally replaces the climb or descent mode with altitude hold.
Does autopilot control airspeed and engine power?
Autopilot can control airspeed with pitch, but engine power is normally controlled by the pilot or a separate autothrottle or autothrust system.
In an airspeed or flight-level-change mode, the autopilot may pitch up or down to maintain the selected speed, allowing the climb or descent rate to vary. In vertical-speed mode, it holds the selected rate instead, so airspeed can rise or decay unless thrust is adjusted.
Managed vertical modes may coordinate pitch and automatic thrust, but the exact priority depends on the aircraft. An autopilot speed target therefore does not prove that power is automatic. See our guide to how automatic thrust control complements the autopilot for the division between pitch and engine control.
Can autopilot take off and land a plane?
A conventional aircraft autopilot does not normally perform the take-off, but a suitably equipped and certified system can fly an automatic approach, flare, touchdown and sometimes runway rollout.
During take-off, pilots control the aircraft while the flight director and, where fitted, automatic thrust provide guidance or power management. Autopilot engagement is permitted only after the aircraft-specific minimum height and other conditions are met. There is no universal engagement altitude; our guide to take-off guidance and minimum autopilot engagement explains why the figure differs between aircraft and operations.
An autoland requires more than selecting approach mode. The aircraft, crew and operation must be approved, the required navigation and radio-altimeter data must be valid, and any required redundant autopilot channels must be available. The runway guidance facilities, wind and system status must also meet the applicable limits.
Approach mode alone does not guarantee an automatic landing. On many flights it captures and tracks the approach before the pilot disconnects and lands manually. Even during autoland, the crew configures the aircraft, verifies speeds and mode changes, monitors the flight path and remains ready to take over; landing gear, flaps, braking and reverse thrust are not simply handled by the autopilot.
Specialised emergency landing systems can choose an airport and land after activation, but those are integrated emergency systems rather than normal autopilot operation.
Why does an autopilot sometimes behave incorrectly?
Unexpected autopilot behaviour is most often caused by the active mode, selected target, navigation source or thrust state differing from what the pilot intended.
| Symptom | Common causes | What to verify |
|---|---|---|
| Autopilot will not engage | Aircraft outside engagement limits, excessive control force, poor trim, invalid sensor data, an active disconnect control or a system fault | Stabilise and trim the aircraft, release control pressure and check warnings and the approved checklist |
| Aircraft turns away from the route | Heading mode still active, wrong navigation source, route discontinuity or unsuitable intercept geometry | Active lateral mode, CDI or navigation source, active route leg and displayed course |
| Aircraft does not climb after a higher altitude is selected | No climb mode has been activated, altitude capture remains active or thrust is insufficient | Vertical mode, selected altitude, thrust and airspeed |
| Aircraft levels earlier than expected | It has captured the selected altitude or a programmed altitude constraint | Altitude selector, FMA and active flight-plan constraint |
| Airspeed rises or decays | Pitch is controlling vertical speed or flight path while thrust remains manual | Pitch mode, thrust mode, power setting and aircraft configuration |
| Autopilot oscillates or disconnects | Turbulence, trim or control-force limits, sensor disagreement or a flight-control fault | Warnings, trim, sensor indications and the aircraft checklist |
If the automation produces an unsafe flight path, the correct priority is the aircraft rather than the panel:
- Maintain control. Disconnect and hand-fly if necessary, then establish a safe attitude, flight path and airspeed.
- Read the FMA. Identify the modes that were active and armed instead of relying on the button that was pressed.
- Check targets and sources. Verify heading, altitude, speed, vertical rate, navigation source and active route leg.
- Check energy and trim. Confirm thrust, airspeed, configuration and whether sustained control pressure or poor trim caused the disconnect.
- Simplify the automation. Use a basic, understood mode only after the aircraft is stable. In a real aircraft, follow the approved abnormal or emergency checklist before re-engaging.
An autopilot is literal: it follows the active command even when that command is unsuitable. It does not know that the pilot meant to select another altitude, expected navigation mode to capture or forgot that thrust was manual.
Does autopilot work the same way in a flight simulator?
Flight-simulator autopilots use the same broad feedback and mode logic, but their accuracy depends on the simulator, aircraft model and avionics implementation.
Common simulator-only problems include duplicate autopilot-disconnect bindings, noisy joystick inputs, hardware controls that disagree with virtual panel positions, an incorrect GPS or radio navigation source and time acceleration that destabilises the simulated control loop. Add-on aircraft may also reproduce engagement limits and mode logic that a simpler default aircraft omits.
When troubleshooting, return to normal simulation rate, stabilise and trim the aircraft, neutralise physical controls, check duplicate assignments and test a basic lateral and vertical hold mode before adding route or approach guidance. The FMA remains more useful than the autopilot master light.
Our detailed explanation of how feedback loops, sensors and flight modes are represented in flight simulators covers those implementation differences. In both simulation and real flying, reliable autopilot use comes from knowing the active mode, its target, its navigation source and who or what is controlling thrust.