Overview - What is Control Theory?

Controls are how we get what we want out of a system (aka plant or process). We have two main types of control: open loop and closed loop.

In an open loop system, we simply write control inputs to our plant and trust that the output is close enough to what we want. This is quite functional for a variety of processes! For example, a toaster is an open loop system: once the desired “done-ness” is set, the toaster will just run for a set amount of time until the bread is toasted (I think). This example also helps illustrate the limits of open loop control— the toaster does the same thing no matter what you’ve stuck in it, so your frozen bread may not come out fully thawed, one type of bread might burn at the same setting, etc.

If we wanted our toaster to be more reliable in the face of the diversity of bread input options, we might opt for a closed loop system, in which we would need to add some type of sensor to evaluate the output of the process (the done-ness of the toast), and then feed that back into the controller. So, if you toast your bread while closely watching it, you are perhaps the closed loop controller of this system.

Source: https://www.researchgate.net/figure/Open-and-Closed-Loop-Systems-Both-can-be-found-in-the-project-management-domain-But_fig2_2537163

For the purposes of this class, it’s worth noting that basically all of the CNC machines you encounter are open loop systems. If your stepper motors are big enough, it’s fairly reasonable to assume that when you tell it to make a step, it makes a step. Because this recitation is on controls and machine learning, we’re not going to spend very much time on open loop controls, but it would be a mistake to think that means that using open loop controls is wrong for you— it’s pretty likely that what you want is an open loop system.

On to the closed loop systems!