Lecture-only students fail more often
A meta-analysis of 225 STEM studies found exam scores rose by 0.47 standard deviations under active learning, and students in traditional lectures were 1.5 times more likely to fail.
Freeman et al., PNAS 2014
Robotics Corner · free interactive courses
Skills you can run,
Skills you can run,
not just recall
Robotics, AI, C++ and security are learned with your hands. Every lesson here turns a concept into something you run, break and fix in your browser — so what you study becomes what you can build, debug and defend.
- 75lessons in three courses
- 160labs that run in your browser
- 400+quiz questions with explanations
- 56hours of guided practice
- $0no fee, no sign-up, no install
The value gap
Watching is not the same as knowing
A video on inverse kinematics feels clear while it plays. A week later, few people can solve one. The gap between “I have seen it” and “I can do it” is where most technical learning is lost — and where these courses are built to work.
Passive course
- Watch a lecture, nod along
- Formulas you copy but never run
- Examples that always work
- A quiz that checks recall of words
- A certificate that proves attendance
These courses
- Read a short concept, then run it immediately
- Neural networks that train live, in seconds
- Labs that let you break things on purpose
- Quizzes that explain why each answer is right
- A challenge per lesson that proves you can apply it
Testing yourself beats re-reading
Students who practised retrieving material remembered substantially more two days and one week later than students who re-studied it — even though re-studying made them feel more confident.
Roediger & Karpicke, Psychological Science 2006
Try it
What stays with you after 30 days?
Switch on the parts of a lesson you actually do. The grey line is reading alone; the coloured line is your study plan. Each time you retrieve or apply an idea, it fades more slowly.
Illustrative model
—
Memory retained over a month
Time invested—
Kept at day 30—
Reading only—
A teaching model built on the forgetting curve: each successful retrieval resets memory and slows the next decay, and the gain is larger when the idea had started to fade. Shapes follow the research above; the exact percentages are illustrative, not measured.
How you learn
One loop, 75 times
Every lesson in every course follows the same five-step loop. It is short enough to finish in an evening and complete enough that the idea sticks.
-
1
Read
A focused concept with numbers, names and diagrams. Formulas and code sit in their own blocks, so the main text stays readable.
-
2
Run
A live lab: an arm solves its kinematics, a policy trains, a packet sniffer fills up — all on your machine, in seconds.
-
3
Break
Move the slider past the safe range. See the robot drift, the pointer dangle, the model overfit. Failure is the fastest teacher.
-
4
Check
Five or six questions that test understanding, not vocabulary. Every answer comes with the reason it is right.
-
5
Build
A practice challenge with a hint and a worked solution note — the step that turns understanding into ability.
What you gain
Abilities you can point to
Each course is built around things you will be able to do afterwards — the kind of abilities that come up in design reviews, interviews and on the factory floor.
Modern Robotics & Physical AI
- Explain how a robot goes from pixels to motor commands — classical and learned
- Compare behavior cloning, diffusion policies, VLAs and RL, and pick one for a task
- Plan the data: teleoperation, simulation, human video, flywheels
- Read a demo video critically and price a deployment
Robot-learning engineer · research engineer · robotics PM
Modern C++
- Read and write C++20 that compiles, with every snippet run through
g++ - Reason about the stack, the heap, pointers and object lifetimes
- Design classes, use polymorphism and templates without surprises
- Replace raw memory handling with the STL and smart pointers
Robotics software · embedded · ROS 2 developer
Cyber Security for Robotics & AI
- Threat-model a robot fleet with the CIA triad
- Trace how an attack chain moves from a web API to firmware
- Harden networks, identities and robot firmware
- Spot adversarial-ML risks and run an incident response
Product security · security engineer · AI safety
Why three courses
A real robot needs all three
A learned policy is useless if its code misses a 10 ms deadline, and dangerous if anyone on the network can send it commands. The library covers the full stack a working product depends on.
Example: the robotics lesson on latency and real-time operation is where C++ memory handling starts to matter, and its safety lesson hands off to robot & firmware security.
Intelligence
Robotics & Physical AI
what the robot decides to do
Execution
Modern C++
how fast and how safely it runs
Trust
Cyber Security
who is allowed to change it
Why now
The robots are already here
Robots are moving from research labs into factories, warehouses and homes, and the people who can build, program and secure them are in short supply.
IFR World Robotics 2026, released 24 Sep 2026~5Mindustrial robots operating in factories worldwide, a record
600k+new industrial robots installed in 2025 alone
+11%growth in annual installations year over year
Your path
Pick a goal, get a plan
You do not need all 75 lessons to get value. Choose where you want to go and you get a focused sequence across the three courses, with real time estimates.
—
—
Who gets value
Different people, different returns
Students
Close the gap between coursework and what robotics teams actually use today — VLAs, diffusion policies, sim-to-real.
Engineers switching
Bring software, ML or embedded experience into robotics without starting from zero or paying for a bootcamp.
Teams & teachers
Use the lessons in class or onboarding: texts are CC BY-NC 4.0 and lab code is MIT-licensed.
Managers & founders
Judge demos and vendors with the lessons on evaluation, the market and deployment economics.
What it costs you
Only your time
- Freeevery lesson, lab and quiz
- Privateno account, cookies or analytics; progress stays in your browser
- No setupno robot, GPU or compiler install needed
- Anywhereworks on a laptop or a phone
Start here
Three courses, one library
Take them in any order. Each one stands on its own and links to the others where they meet.
Modern Robotics & Physical AI
From classical control and behavior cloning to VLAs, humanoids, world models and deployment.
Open the courseModern C++
The language robots run on: pointers, the heap, classes, polymorphism, templates, the STL and smart pointers.
Open the courseCyber Security for Robotics & AI
Think like a defender: threat models, attack chains, firmware and network defence, adversarial ML, incident response.
Open the course
FAQ
Questions & answers
Is it really free?
Yes. All three courses are free with no sign-up. Texts are released under CC BY-NC 4.0 and lab code under the MIT license.
Do I need a robot or a powerful computer?
No. Every lab runs in the browser — networks are small enough to train in seconds, and simulations are built for a laptop or a phone.
Which course should I start with?
If you are new to programming, start with Modern C++. If you can already code and want robotics, start with lesson 0.1 of Modern Robotics. If you work on products, the security course reads well on its own. The path builder above gives a concrete order.
What background do I need?
For robotics: some Python and the basics of machine learning. The C++ course starts from zero. The security course assumes general computer literacy.
Will this get me a job?
No course can promise that. What these courses give you is the ability to explain, run and reason about the systems employers build — and practice challenges you can extend into portfolio projects.
Is any of my data collected?
No. There are no accounts, no cookies and no analytics. Your progress is stored only in your own browser.
Your first lab is one click away
Stop watching.
Stop watching.
Start building.
Pick a course and run your first lab in the next five minutes.
Free · No sign-up · Labs run in your browser