In our last session, we discovered that you’re already a natural at control theory. You do it every time you balance a broom, adjust the shower temperature, or ride a bike. We boiled down this complex-sounding field to a simple, four-step dance: Measure, Compare, Compute, and Correct. This cycle, known as a feedback loop, is the “secret brain” that allows systems to adapt and achieve their goals.
But what if a system skipped a few steps? What if it just followed orders without ever checking its work?
As it turns out, not all control systems are built the same. They fall into two major families, each with a fundamentally different philosophy. One is the trusting, fire-and-forget type, and the other is the skeptical, constantly-verifying type. Understanding the difference between them is the next crucial step on our journey. Let’s meet the two fundamental approaches to control: Open-Loop and Closed-Loop.
The Optimist: Open-Loop Control
Imagine you’re using a basic, old-school toaster. You put your bread in, turn the dial to “3,” and push the lever down. A timer starts ticking. After a few minutes, your toast pops up. The toaster did its job. It followed your command. 1
But what if you were toasting a thinner slice of bread? It might come out burnt. What if the bread was frozen? It might still be cold in the middle. The toaster doesn’t know and doesn’t care. It was given one command—”run for the amount of time that ‘3’ means”—and it executed that command blindly.
This is open-loop control. It’s a one-way street. A command goes in, an action comes out, and that’s the end of the story. There is no feedback. The system has no way of knowing the actual result of its actions, so it can’t make corrections. 3
You see open-loop systems everywhere because they are simple and cheap.1
- A clothes dryer on a timer: It runs for 60 minutes, regardless of whether your clothes were dry in 30 or are still damp at the end.6
- A simple garden sprinkler on a schedule: It turns on at 6 AM every day, even if it’s pouring rain.6
- A traffic light on a fixed timer: It cycles through red, yellow, and green at the same pace, even if there’s a huge traffic jam in one direction and no cars in the other.5
Open-loop systems work perfectly fine when the world is predictable and you don’t need high precision. You trust that turning the dial to ‘3’ will give you decent toast most of the time. But when the unexpected happens—a change in the bread, a sudden rainstorm, a traffic surge—the system fails to adapt because it’s flying blind.
The Realist: Closed-Loop Control
Now, let’s upgrade our toaster. This is a “smart” toaster. It has a sensor inside that measures the color of the bread. Your command is no longer “toast for three minutes,” but “toast until golden brown.”
You put the bread in and press start. The heating elements turn on, but this time, the toaster is watching. The sensor constantly feeds information about the bread’s color back to the toaster’s tiny computer (the controller). The controller compares the current color to the “golden brown” color you requested. As the toast gets closer to the goal, the controller might even reduce the heat to avoid overshooting. Once the sensor reports that the goal is reached, the controller shuts everything off and pops up your perfect toast.
This is closed-loop control, and it should sound familiar. It’s the full Measure, Compare, Compute, Correct loop in action.6 The key ingredient is feedback. The system uses a sensor to measure the actual output, allowing it to adapt to changing conditions and actively work to eliminate any error between the goal and the reality.2
This is the type of control that enables intelligence and precision:
- A car’s cruise control: It doesn’t just lock the gas pedal. It constantly measures the car’s speed and adjusts the throttle to maintain that speed whether you’re going uphill or downhill.7
- A home thermostat: It measures the room’s temperature and turns the heat or AC on and off to keep it exactly where you want it.2
- Your own body: When you reach for a glass of water, your eyes are constantly telling your brain where your hand is relative to the glass, allowing you to make millions of tiny corrections for a smooth, successful grab.
Closed-loop systems are more complex and expensive, but they are powerful. They can handle uncertainty and disturbances, making them far more accurate and reliable.2
| Feature | Open-Loop Control | Closed-Loop Control |
| Feedback | None. It’s a non-feedback system.5 | Yes. It uses feedback to self-correct.5 |
| Key Components | Controller, Actuator.9 | Controller, Actuator, Sensor, Comparator.9 |
| Complexity | Simple and easy to build.1 | More complex and difficult to design.5 |
| Cost | Inexpensive.1 | More expensive.5 |
| Accuracy | Less accurate; depends on calibration.5 | Highly accurate; constantly adjusts for errors.4 |
| Reliability | Unreliable when conditions change.1 | More reliable and can adapt to disturbances.9 |
| Example | A timed microwave. | A thermostat-controlled air conditioner.5 |
Tying It to Our Robot
So, how does this apply to the mecanum wheel robot we’re going to build?
We could try an open-loop approach. To move forward one meter, we could calculate that the motors need to spin at a certain power for, say, two seconds. We could send that command and just trust that it works. But what if the battery is low, making the motors weaker? What if the floor is extra slippery? What if one wheel has slightly more friction than the others? The robot wouldn’t move one meter, and it probably wouldn’t even go straight. It would fail because it trusted its commands without verifying the results.
Instead, we will build a closed-loop system. Our robot will have sensors on each wheel called encoders that measure how fast each wheel is actually spinning.10
- Measure: The encoders will constantly report the real-time speed of each wheel.
- Compare: Our controller (the robot’s brain) will compare this actual speed to the target speed required for the desired movement (e.g., moving straight forward at 0.5 meters per second).
- Compute: The controller will calculate the error for each wheel and decide how to fix it. If a wheel is too slow, it needs more power. If it’s too fast, it needs less.
- Correct: The controller will adjust the power sent to each motor individually.
This loop will run hundreds of times per second. By constantly checking its work and correcting errors, our robot will be able to drive with incredible precision, compensating for a slippery floor or a dying battery in real-time. This is the power of closing the loop.
What’s Next?
We’ve established that for any task requiring precision and adaptability, a closed-loop system is the way to go. But that leaves a big question: what exactly happens inside the “Compute” step? How does the controller’s brain actually decide how to correct an error?
In our next session, we’ll open up that black box. We’ll meet the three core components that make up the most famous and effective controller in the world: the Proportional, the Integral, and the Derivative terms. Get ready to meet the PID controller.
