Open Loop vs. Closed Loop

Imagine throwing a dart while wearing a blindfold. You know roughly where the board is, you feel the dart in your hand, and you throw with the force you think is right. You have no way to see where the dart lands or correct your throw mid-flight. That is open-loop control. Now remove the blindfold. You watch the dart leave your hand, see where it hits, adjust your stance and force, and throw again. Each throw is informed by the result of the last one. That is closed-loop control. The difference is whether the system uses information about its own output to guide future actions.

Open-Loop Control: Commanding Without Checking

An open-loop system sends a command and moves on. There is no sensor checking whether the command produced the intended result. The controller simply trusts that the actuator did what it was told. A microwave oven is a classic open-loop system. You set 90 seconds, press start, and the microwave runs for 90 seconds. It has no idea whether your food is hot, warm, or still frozen. It cannot adjust. When 90 seconds elapse, it stops — regardless of outcome. A simple robot arm running open-loop is similar: the controller sends a pulse that should move the joint to 45 degrees, waits the expected time, and proceeds to the next step. If friction slowed the joint, if the power supply dipped slightly, or if a bump knocked the arm mid-movement, the controller has no idea. It assumes success whether or not the joint actually reached 45 degrees. Open-loop control is not useless. It works well when the system is reliable, the environment is predictable, and errors accumulate slowly. Stepper motors in 3D printers often run open-loop — the mechanics are precise enough that missed steps are rare under normal operating conditions.

Closed-Loop Control: Sensing and Correcting

A closed-loop system adds a feedback path: a sensor measures the actual output, the controller compares the measured output to the desired output, and the difference — called the error — is used to compute a corrective command. The loop is closed because information flows from output back to input, forming a circle. A household thermostat is a simple closed-loop controller. The desired output is your set temperature — say 22 degrees Celsius. The sensor measures the actual room temperature. If actual is below desired, the furnace turns on. When actual reaches 22, the furnace turns off. The system continuously checks and corrects. A servo motor joint works the same way at high speed. The desired angle is 45 degrees. The sensor reads the actual angle. If it reads 43 degrees, the motor drives toward 45. If it reads 46, the motor drives back. This happens hundreds of times per second, keeping the joint tightly locked to the target. The error signal — desired minus actual — is the core of all closed-loop control. When error is zero, the system is exactly where it should be.

Why Almost Every Precise Robot Uses Closed-Loop

The real world is full of disturbances: friction that varies with temperature, loads that change as an arm picks up objects of different weights, vibrations from nearby machinery, and small manufacturing imperfections in joints and gears. An open-loop controller ignores all of these. A closed-loop controller automatically compensates for them, because any disturbance that pushes the output away from the target immediately shows up as an error and triggers a correction. This is why surgical robots, industrial assembly arms, self-driving cars, drone stabilizers, and computer hard-drive read heads all use closed-loop control. The cost is added complexity: you need sensors, sensor calibration, and a control algorithm smart enough to compute good corrections without overshooting or oscillating.