Why don't CS students learn about control theory?

Imagine that all the math and physics you've learned in you life is Pinocchio from that old Disney movie. Control theory is the magic wand that makes him a real boy.

Feedback loops, gain, stability, blablabla.

Control theory isn't that
You'd probably say that about whatever the first higher level math class you took was about

To be fair, a significant amount of what Boston Dynamics is doing is just developing control systems.

See this user,
basically it's about how to get a system of defined inputs and outputs to do what you want. Shows up a lot in robotics and engineering disciplines. There are some good YouTube lectures on it.

Probably a similar reason as why CS students don't learn about fluid dynamics

So CS students are supposed to be incapable of doing any interacting with physical systems?

Yes it's a theory degree, CS is basically a math degree that has some programming on the side because it's a common tool.
I'm sure some intro to robotics course touches on it but it's definitely not that relevant.

This is not to say that the knowledge isn't good or useful, by the way

CS students can barely pass calculus, do you really think they could pass control theory?

Because they're too stupid and bad at math.

>CS is basically a math degree
Chemistry is more of a math degree than CS. Stop spreading that lie.

Attached: Hardest class in CS.png (490x684, 423K)

block diagrams and vector calculus. A lot of that shit with convoluted tricks which the professor will try to catch you out on in your final exam which is worth 80% of the class. Yes I still remember that class 5 years later.

We do, very briefly in a class on computer architecture and organization. Everyone fails that class though. Not me though, because I am a genius.

Who is Alan Turing

a dead faggot

and don't forget his most important achievement.. he was chemically castrated

Not someone who's teaching CS babbies calculus

>There are some good YouTube lectures on it.
youtube.com/watch?v=FurC2unHeXI
MATLAB has us covered, senpai.

Because control theory is more of an engineering topic since it dealt with actual working systems. That said, it depends more on the view of the Uni on how the degree should be taught.

>do Software Engineering
>lots of formal methods
>all nice and dandy
>"so, how do we translate the model to the system?"
>"by hand, user, so don't make mistakes when translating the model"

Not saying it's useless, but certainly designed for people with inclination for safety critical systems or designing electronics.

also

>seen too many o'reilly memes
>tfw you can't tell whether it's a book or a meme at first glance
welp

Attached: javascript-for-millennials.png (515x656, 172K)

because it is a legit academic topic requiring intellectual capacity 2 orders of magnitude higher than cslets possess

Anything from keeping an inverted pendulum steady to your car's cruise control.

Good links:
ctms.engin.umich.edu/CTMS/index.php?aux=Home
see.stanford.edu/Course/EE263

> ML is basically mostly that
Absolutely not.

Maybe, just maybe, if you want to find something in common you could think
of model identification with stochastic model but even there the basic assumptions are orthogonal

ML: y(t) = f(X) + e(t)
CT: y(t + dx) = f(y(t), e(t)) with e(t) ~ WN(u, k)

> if you didn't do continuous systems you didn't learn control theory

Thats bullshit, discrete time control isn't just petri net.

In real world for black box identification and adaptive control you use LTI stochastic discrete time model

Some M.S had Control Theory as requisite but average C.S grade lack good math level, note Engineering learn baby TC because lack Analysis and Modern algebra education.

Yes and in most cases they are.

We do. It’s called something else at some unis but pretty sure everyone has this course

CS majors, what do you fellas learn in control and systems? PID? Lead lag controllers?

Attached: ee0a329c935160e6b576d94f01659e73e648325ca3ec5aadf891486e6b0b4250.jpg (359x347, 22K)

CS majors don't learn that because their degree is in theoretical math and not some random discipline of engineering

PID

>some random discipline of engineering
dude the most applicable thing to every field in engg is systems and control theory. only engg discipline that might not need it is civil i think.

Yes that's why I said "some random discipline of engineering," if you picked one at random they would be learning it.
CS does not qualify as some random discipline of engineering because even if you were desperate to consider it one it would be very far removed.

ah my bad.

CS is fucked because nobody tells all the fags that hear about it and say 'imma make a video game" or "imma make a robot" or "imma be a uml writing code monkey" that it's actually something completely different.
Then they figure out they won't be learning control theory and it's too late.

>ML is basically mostly that.

Lmao... fuck off kid

CS students should take it as an elective instead of intro to Spanish or some other gay shit they’ll just immediately forget.

liberal arts should just be thrown out of campuses altogether. it's all just genderqueer theory shit at the end of the day.

simplest way to think of it:

A motor lets you control the speed by increasing the magnitude of the current you apply to it. You can reverse the flow of the current to make it spin the other way.

This is fine if you're building a model car but what if you wanted to use that motor to build an elevator?

You are now interested in controlling position not speed, but speed is the only way your signals (in this case the current) can interact with the process (in this case, the motor). And the response is now position not speed.

So the simplest introduction to solving this is taking the difference between where you are and where you want to go and multiplying it by some factor and using that to control your current.

It turns out that if you make that factor both negative and smaller than 1 you get something called negative feedback, which results in your elevator eventually getting to your desired position.

( I am glossing over so fucking much right now and I encourage you to actually look it up if you're interested this is just one of many many different control types and the math used to describe them involve heavy calculus)

Do any of you guys use ROS? My professor has me implementing PID control for an autonomous vehicle and I need better example documentation than the wiki for it.

Yes, they are geared towards applications and software development where there is a ton of code and not a lot of math, and a high tolerance for mistakes.

Not control systems where theres often not much code, a ton of math and a small error in the control strategies selected could result in catastrophic failure.

This may sound harsh against software devs, but trust me you do not want any of these control system mother fuckers anywhere near your codebase, they do not understand how to program despite what they think.