ublo
bogdan's (micro)blog

bogdan » academia is failing robotics (and here’s why)

10:15 pm on Oct 13, 2025 | read the article | tags: buggy

universities are still teaching robotics like it’s 1984. meanwhile, the world’s being rebuilt by people who just build things.

i came across a «robotics» course proudly adding material science labs, as if that’s what makes robots move. i almost laughed.
because that’s the problem with academia today, it’s completely disconnected from reality.

the ivory tower illusion.

universities still think theory builds engineers. they believe if you can recite equations, you can build robots.

wrong.

in the real world, things need to work, not just make sense on a whiteboard.

when i worked for this German automaker predicting car-part failures, we had 20+ people in data science. only one (a part-time veteran with 30 years in the industry) had ever worked on actual cars.

guess who understood the problem?
yeah. the one who’d fixed real engines.

teaching for the 5%.

universities are built for the top 5%: future professors, paper writers, grant chasers.

the other 95%, the people who could build the next Tesla, Boston Dynamics, or SpaceX, are being buried under irrelevant abstractions taught by people who’ve never touched hardware. it’s an education system optimized for theoretical superiority, not functional competence.

if you want robotics, start with code!

forget the buzzwords.

$$robotics = coding + electronics + feedback \: loops$$

you need:

• python: your prototyping glue and AI backbone;
• c or rust: your low-level control, where timing and efficiency matter;
• basic hardware & network intuition: how computers talk, how currents flow, how systems fail.

no one cares how silicon is doped. but everyone cares why your MOSFET circuit leaks current when driving an LED array (spoiler: reverse breakdown. you learn that the hard way.)

that’s real engineering. not the fantasy in textbooks.

engineering ≠ science.

science explains the world. engineering builds it. academia keeps confusing the two: creating “theoreticians of machines” instead of builders of machines.

in labs, they avoid failure. in industry, we learn from it. that’s why SpaceX blows up rockets on purpose. and moves faster than entire university departments writing papers about «optimal design».

the revolution starts small:

that’s why the AI & IoT club i’m starting for physics students won’t look like a class. no 200-slide powerpoints. no «history of transistors 101». we’ll build something every second week.

small, vertical slices: sensors → code → network → actuation. things that work.

if it fails, perfect. we debug. that’s called learning.

we’ll invite engineers, not theorists. we’ll publish open-source projects, not academic reports. we’ll measure progress in blinking LEDs, not credits earned.

stop pretending. start building!

the future of robotics, IoT, and AI will belong to those who can code, connect, and iterate. not those who can only talk about it.

academia can keep polishing its powerpoints. we’ll be busy making the future boot up.