Doing a school project on Industry 4.0 is a waste of time. I don't know what a magister is. Is similar to what we call a master's degree? I don't see how an automation degree is useful since automation is really knowing how to control machinery and there are many types of machines. PH control required knowledge of chemistry and how ions associate and disassociate. Temperature control requires knowing how to compensate for dead time. Each application requires knowledge of more than just flipping bits.
Worth while topics of study.
Learning how to do system identification and modeling. Model the systems using differential equations.
Different types of filtering or noise reduction. One can't control what what can't measure. It is often necessary to filter or estimate actual values. This is somewhat related to the previous line.
For motion control a big topic is target generators. There are the simpler one dimensional target generators but robotics require 3 dimensional motion generates. The math is complicated but it is also already done.
Generate all the formulas for calculating controller gains. Include pole AND zero placement.
What can one do with zero and pole placement?
Do this for up to 3rd order systems. SOPDT has been done. Why doesn't one hear of TOPDT plants?
How does one optimize a system that requires cascaded control. When is cascaded control required?
What is input shaping? This is an interesting topic. I got introduced in the early 1990s.
Here is the problem from hell. So far I know of only myself and one other person has be able to solve it. Move from any position, velocity and acceleration to any other position velocity acceleration in the minimum time given a jerk limit. A professor used my problem in a seminar many years ago. His software would solve the problem but not fast enough so it could be used in a motion controller.
Students often have projects where a car needs to follow a line around a track. Although some teams get very good results, the project is much more difficult. It requires using the chain rule because the PID should not be updated as a function of time. These kinds of problems have shown on this forum a few times.
How about optimizing multiple input and multiple output problems.
Model predictive control is also a good topic. The MPC algorithm is probably not as difficult as calculating the model to use. This is related to above suggestion.
Being able to model and optimize things will go a long way. It is way beyond turning things off and on.
This is one of my favorite quotes by a dead white dude.
William Thompson AKA Lord Kelvin said:
I often say that when you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meagre and unsatisfactory kind.
I really like the idea of modeling and optimizing.