What is the Reference in Control Theory?

I've started studying Control and I've come across some concepts that I'm finding a bit difficult to understand.

For example, in the system showed here, there is a signal called "reference" as well as a feedback block. As far as I know, the feedback block has the function of getting the output as close as possible to the reference. But here is what I don't understand: what if the reference is, for example, an impulse, and I want the system to show a step when an impulse is at its input?

I'm getting confused with the difference between reference signal and input signal. Why would someone connect a feedback block in order to get the output to follow the reference? What's the advantage of doing this?