Autopilot

An autopilot is an advanced system used to control an aircraft's flight path.

It does not replace the human pilot. Instead, it assists them by handling repetitive tasks. This allows pilots to focus on broader operations like monitoring weather or on-board systems.

Autopilots often work with an autothrottle. This is a system that controls the power delivered by the engines. Together, they manage the aircraft's speed and trajectory.

Modern autopilots rely on complex computer software to guide the aircraft. The software reads the plane's current position and attitude. It then instructs the flight control system to make necessary adjustments.

Large aircraft typically use an inertial guidance system to track their movement. However, these systems can experience drift, which is a corruption of positional data. To fix this, the system uses digital signal processing, often a six-dimensional Kalman filter. This filter resolves disagreements across roll, pitch, yaw, altitude, latitude, and longitude. Pilots also use radio aids like GPS or DME to correct the aircraft's position.

Autopilots are categorized by how many axes they control. A single-axis autopilot only manages the roll axis. These are sometimes called "wing levellers" because they keep the wings level. A two-axis autopilot adds control over the pitch axis, which is the up and down movement. This allows for automatic flight guidance after takeoff. A three-axis autopilot adds control over the yaw axis, which is the side-to-side movement. Most modern complex aircraft use three-axis systems to manage all flight phases. These phases include taxi, takeoff, climb, cruise, descent, approach, and landing.

Some systems are designed for specific, difficult tasks. An Autoland is a system that can land a plane on a runway. It uses an Instrument Landing System (ILS) Cat IIIc approach. This is used when visibility is zero, such as during thick fog. The aircraft can often stop on its own during an Autoland. However, a pilot must disengage the autopilot to taxi to the gate. Another option is Control Wheel Steering (CWS). In CWS mode, the pilot uses the yoke or stick to provide inputs. The autopilot then holds that specific heading and attitude to provide stability.

The history of this technology began in the early 1900s. In 1912, the Sperry Corporation developed the first gyroscopic autopilot. It used a gyroscopic heading indicator and attitude indicator. These were connected to hydraulically operated elevators and rudders. In 1914, Lawrence Sperry demonstrated the system in Paris. He showed it worked by flying with his hands away from the controls. In 1930, a more compact version kept a U.S. Army Air Corps aircraft on course for three hours. By 1947, a C-53 aircraft completed a transatlantic flight using an autopilot for takeoff and landing.

Safety is the most important part of autopilot design. Engineers use redundancy to ensure the system keeps working if a part fails. For example, the Boeing 777 uses triplicated microprocessors. These are fabricated in a radiation-resistant process. Some systems also use design diversity to prevent software errors. This means different engineering teams write the software using different programming languages. The Space Shuttle used this method with five computers. Four ran identical software, while a fifth backup ran independent software. This reduced the chance of a shared mistake causing a failure.

Another important system is the stability augmentation system (SAS). Unlike a standard autopilot, an SAS does not maintain a specific altitude or path. Instead, it moves control surfaces to damp out unwanted motions. A common type is the yaw damper, which helps swept-wing aircraft. It uses sensors to detect the start of a "Dutch roll" instability. A computer then tells an actuator to move the rudder in the opposite direction. This stabilizes the aircraft about the yaw axis. Some aircraft, like the B-52, require both pitch and yaw SAS to remain stable.