feedback control

Feedback control means a machine keeps watching what it is actually doing, compares that to what it was told to do, and constantly nudges itself to close the gap. The everyday picture is steering a car: you do not turn the wheel once to a fixed angle and hope; you keep looking at the road, notice you are drifting left, and ease the wheel right — over and over, many times a second, without even thinking. The measurement (where the car is) flows back to your hands (what they do), which is exactly why it is called feedback.

The trick that makes this powerful is that the system corrects for surprises it was never told about. A heating thermostat does not need to know that you opened a window, that a cold front blew in, or that ten guests just walked in radiating warmth — it only needs to feel the room temperature, see that it has fallen below the target, and turn the heat up until the reading climbs back. Because it reacts to the actual result rather than to a fixed plan, feedback control stays accurate even when the world misbehaves, parts wear out, or the load changes.

The cost of all this watching is that feedback can overshoot, wobble, or even shake itself apart if it reacts too hard or too late — like over-correcting the steering wheel and weaving down the road. Most of the craft of classical control is about choosing how strongly and how quickly to respond so the machine settles smoothly on its goal instead of fighting itself.