First, think about what a pilot does when the aircraft gets disturbed — Page 358, Lesson 420

Let me walk you through what the autopilot is actually doing here, because this is the heart of how a basic autopilot works. First, think about what a pilot does when the aircraft gets disturbed. Say a gust of wind pushes the nose up. The pilot senses that change in attitude, works out how much and in which direction to move the controls to correct it, and applies that correction with his muscles. The aircraft then manoeuvres about its centre of gravity back towards its original position. Once he senses that the correction has taken effect, he moves the flight controls back to neutral to remove the correcting input. Now here's the key idea: the autopilot does exactly the same job, but in a shorter reaction time. It must detect the disturbance, work out the appropriate corrective action, apply that correction using its "muscles" — which causes the aircraft to manoeuvre about its centre of gravity back towards its original position — and finally detect that the correction has taken place and re-centre the controls. This whole process is called inner loop control, sometimes called closed loop or auto-stabilization. Let me break that down into the five steps the autopilot performs, in list form: 1. Sense changes in attitude — it detects that the aircraft has been disturbed. 2. Compute the amount and direction of control required — it works out the correction. 3. Provide the muscle to move the control surfaces using servomotors — the servomotors are the autopilot's muscles; they physically move the control surfaces. 4. Detect that the control has been applied and that the aircraft has responded — it checks that the correction actually worked. 5. Return control surfaces back to the neutral condition when the disturbance has been corrected. Now, you might be thinking: aircraft are naturally stable anyway, so why bother installing this basic autopilot at all? That's a fair question. Yes, aircraft are naturally stable, but there are different types of stability, and that stability can change with respect to the ambient conditions — things like air density, temperature, or speed. The basic autopilot augments and fine-tunes that natural stability, providing enhanced stability over a greater range of ambient conditions. One thing I want to emphasize: the most basic autopilot available will only provide auto-stabilization. That's its entire job — it's not doing navigation or anything fancy. It's purely there to maintain stability. Let me show you this visually. So the takeaway here is the loop: sense, compute, act, verify, re-centre. That's the inner loop, and it's the foundation everything else in autopilot design builds on.