Free intro course
Robotics University: the intro course
A hands-on survey course on robotics in plain English — every idea explained with words you already know, a short video worth watching, and a robot you can actually move. For self-motivated beginners and career switchers. No hardware, no grades, no jargon.
The order below is a suggestion. Each module says what it assumes; nothing is locked, and you can start anywhere.
- 42 modules
- 7 robots
- 7 parts
- 0 prerequisites
- about 9 min each
Part 0
What a robot is
What makes something a robot rather than a machine, what it is built from, and what "it can move six ways" actually means.
- Module 01What makes something a robot?A machine that notices, decides and does — over and over. Switch each one off and see what is left.
- Module 02The parts of a robotLinks, joints, motors, sensors, and a computer — find each one on a real arm.
- Module 03Degrees of freedomHow many different ways can this thing move?
Part 1
How robots move
Joints, reach, motors and wheels. Where the hand is, how to get it somewhere, and why arms sag and rovers turn without a steering wheel.
- Module 04Joints: the ones that turn and the ones that slideNearly every joint on every robot is a door or a drawer. Move five of them on their own, and find their limits.
- Module 05Where is the hand? Forward kinematicsTell a robot its joint angles and it can tell you, without looking, exactly where its hand is. Do the sum yourself and put the hand on a star.
- Module 06Frames: where "left" is depends on who you askA position is always counted from somewhere, facing some way. Watch one star get three different addresses, and see why a camera on the hand needs the base's help.
- Module 07Getting the hand there: inverse kinematicsPoint at a spot and the robot has to find the joint angles that put its hand there. Sometimes there are two ways to do it, sometimes none. Drag a star and watch it work.
- Module 08Reach: where the hand can and cannot goEvery robot arm is surrounded by an invisible shape, the set of all places its hand can be. Sweep a star through the air around a real arm and paint that shape.
- Module 09Motors: force, speed, and gearsA motor is strongest standing still and weakest at full speed. Put a weight in a real arm's hand, shift its gears, and watch it choose between fast and strong.
- Module 10Gravity and momentum: why arms sag and overshootAn arm told to hold still sits a little low, and an arm told to stop carries on past. Put a weight in a real arm's hand, then turn gravity off, and find out which of the two goes away.
- Module 11Grippers and hands: holding onA hand holds on with two things, squeeze and friction, and either can run out. Watch a real arm's jaws pick up a block, then make the block too slippery, then the jaws too weak, and see which fails how.
- Module 12Wheels: how rovers turn with no steering wheelA rover has two wheels, two motors and nothing that steers. Drive one, watch what each wheel is told, and see that a difference between the wheels is what turns it.
- Module 13Legs and balance: why walking is hardA rover cannot fall over. A dog can, every step it takes. Walk a four-legged robot, shove it, and see what four stiff legs can hold and what they cannot.
- Module 14Flying: four motors and a lot of math the robot does for youA quadrotor cannot stay up for a moment without a controller correcting it hundreds of times a second. Fly a palm-sized drone, watch what its four motors are doing, and then take one of them away.
Part 2
How robots sense
Knowing where your own joints are, feeling a grip, measuring distance, seeing, and being wrong about all of it by a little.
- Module 15Knowing where your own joints are: encodersEvery joint on the arm reports its angle as a whole number, and those numbers are all the robot has. See where the six encoders sit, read what they report, and find out what they cannot tell you.
- Module 16Touch and force: feeling the worldThe arm's jaws are around an egg, and a panel reports every push the arm feels, in newtons. Close the jaws a half degree at a time, watch the squeeze climb, and find the step that cracks it.
- Module 17Measuring distance: sonar and lidarFive beams on the rover's nose, each reporting one number, the distance to the first thing in its way. Watch a box walk from beam to beam as the rover turns past it, and find the thing twenty centimeters in front of the nose that no beam can see.
- Module 18Cameras: what a robot actually seesThe arm's wrist camera hands the robot a grid of colored squares with a block somewhere in it. Watch the block go from four blue squares to a cube with edges, and see what no picture tells the robot: how far away anything is.
- Module 19Odometry: where am I, and why wheels lieA rover with its eyes shut can still say where it is, by counting what its wheels have done. Watch its belief ride on the truth for a whole square, then oil one wheel and watch the two come apart by a meter.
- Module 20Noise: why every sensor lies a little, and how averaging helpsTwo rovers with the same shaky sensors drive at the same box. One believes every reading and flinches at a box that is not there yet; the other believes the average of the last few, and drives smoothly. Then make it average too much, and watch it turn too late.
Part 3
How robots are controlled
Feedback: noticing the gap between what you wanted and what happened, and doing something sensible about it without shaking yourself apart.
- Module 21Open loop, closed loop: the idea of feedbackTwo rovers set off for the same beacon. One drives straight ahead for three seconds, because that is where the beacon was when someone measured. The other keeps looking at the beacon and steers toward it. Put oil under a wheel and only one of them still arrives.
- Module 22Proportional control: push harder the further you areTwo copies of the same arm run the same simple rule: push harder the further you are from the target. The only difference is how hard, and that one number decides how far each arm sags, how far a knock sends it, and how it shivers back.
- Module 23Push, patience and a brake: what PID stands forThe rule inside almost every robot joint is called PID, three letters for three pushes added together. You met the first one last time. This module adds the other two, one switch at a time, and shows what each one does to an arm lifting a weight.
- Module 24Stability: when control goes wrongA cart balances a pole, and one slider delays the news of the pole leaning. Fifty milliseconds late and the pole rocks for ages; sixty, and every swing is bigger than the last until it falls. Nothing changed but when the news arrives.
- Module 25Trajectories: getting from A to B smoothlyEvery star you dragged so far was you planning for the robot. Left to plan for itself, a robot's plainest plan, straight lines at one speed, asks its motors for the impossible at every corner. Two arms run the same three points, one jerky and one smooth.
- Module 26Balancing something that wants to fallThe cart-pole is back, and the four pushes that keep its pole up are yours to switch off. Each one missing brings the pole down its own way: straight over, whipped over, walked off the rail, or swayed into a bumper.
- Module 27Stiff or soft: holding an egg without breaking itThe spring rule from the proportional arm is aimed at an egg, with its stiffness on a slider. Stiff, the jaw shuts in a blink and cracks the egg; soft, it creeps closed and holds it; too soft, and it never arrives.
Part 4
How robots think
Maps, knowing where you are on one, planning a way through, and deciding what to do next.
- Module 28Maps, and knowing where you are on oneThe rover that counted its wheels and drifted gets a floor plan, and the habit of checking its rangefinders against it. Two thousand guesses are scattered over a small house; watch them die where the walls look wrong and pile up on the rover.
- Module 29Planning a path around obstaclesThe rover knows where it is on the plan of the house and wants the room below, through the wall, with the only door down the corridor. Watch a search spread one cell at a time and find the way, then wall the door up.
- Module 30Dodging things on the flyNo map this time, and no plan. A box slides along the rover's line to the beacon, and the rover gets there by one rule run fifty times a second: everything near pushes the steering away, the beacon pulls it back.
- Module 31Planning for arms: moving without hitting yourselfAn arm knows the angles that put its hand on a star, but a wall stands between, and turning straight to them swings the arm through it. The robot maps the wall in its own joint angles, searches the map, and lifts its hand over.
- Module 32Deciding what to do: state machinesThe robot vacuum from the first module is back, looking for its dock. Its whole brain is five circles and seven arrows, and you can edit them: take one arrow away and it runs on its bumper; swap left for right and it circles for ever.
Part 5
How robots are programmed
Writing the loop yourself, how real robot software is put together, and the ways a simulation lies to you.
- Module 33Telling a robot what to do: commands, loops, timingThe arm is back with no sliders under it, only a program: eleven lines that lift it, wait, wave three times and report where the shoulder ended up. Read it, run it, change a number, and ask for an angle it cannot reach.
- Module 34Writing the sense-think-act loop yourselfThe first module called the cycle behind a vacuum meeting a chair leg the sense-think-act loop. This module writes it: thirteen lines that drive the rover and, five times a second, read the front rangefinder and turn if something is under forty centimeters.
- Module 35How real robot software is organized: nodes and messagesA real robot's software is dozens of small programs, each doing one job at its own pace and passing what it finds to the others by name. Here the rover's is drawn that way, and you can stop any piece and watch what goes quiet.
- Module 36Simulation vs reality: why sim is used, and how it liesTwo arms run one plan: reality on the left, the simulation on the right, its model of the block too grippy and too light. One drops the block, one puts it in the bin. Put the model's errors right one at a time and find which lie mattered.
- Module 37Safety: limits, stops, and why robots are slow near peopleA desk-sized arm swings its hand across the table with a person standing in the way, and hits them at full speed. Switch on a limit, a speed limit and a stop on touch, and find what each one saves and what it costs.
Part 6
How robots learn
Teaching by showing, learning by trial and error, recognizing what a camera sees, and what language models can and cannot do for a robot today.
- Module 38Teaching by showing: imitation learningDrag a robot arm's hand to a dot a few times, then move the dot and let the arm try by itself. It lands on dots between the ones it was shown and misses the ones beyond. Find out why, then fix a miss.
- Module 39Learning by trial and error: reinforcement learningNobody tells this cart how to keep its pole up. It is paid points, tries, fails, and tries again, hundreds of times while you watch. Then change what it is paid for, and watch it learn exactly that instead.
- Module 40Seeing and recognizing: neural networks for perceptionThe rover's camera hands 576 numbers to a small neural network, which answers with a word and how sure it is. Turn the rover to face a block up close, a block far off, and a crate it was never shown.
- Module 41Robots that understand language, and what is real todayType pick up the block and put it in the bin, and a robot arm does it. Say the same thing another way and it is lost. Ask it to pick up an egg and it knows every word, which does the egg no good.
- Module 42Where robotics is going, and where you go nextEvery robot you have driven is here in one box, and nothing is taught. Drive the rover with no brain, cut a flying robot's motor, knock a pole too hard, and find where each one gives out. Then where robotics is going, and what to read and do next.