Tell a robot to drive forward for 3 seconds. On carpet it goes 80 cm; on wood, 1.2 m; with a weak battery, 60 cm. It has no idea. That is open-loop control.
Key idea
Open loop: send a fixed command and hope. Closed loop (feedback): measure the result, compare it with the goal, and correct. The error is $e = \text{setpoint} - \text{measurement}$.
Why it works
Feedback works without knowing the world exactly. Whether the floor is carpet or wood, the robot keeps measuring and correcting until the error is small. That is why almost every robot joint and wheel runs a feedback loop.
Worked example: The error
Goal: 90°. The arm's sensor reads 72°.
- $e = 90 - 72 = 18^\circ$.
- Positive: push toward larger angles.
Answer: 18°.
Worked example: Open or closed?
A thermostat that turns the heater on when the room is below the set temperature.
- It measures the result (room temperature) and corrects.
Answer: Closed loop.
Common mistake
Getting the sign of the error backward (measurement minus setpoint). Then the controller pushes the wrong way and the error grows.