Robotics

A robot, for the purposes of a hobby, is a machine that senses something, decides something and moves. Remove the sensing and you have a remote-control vehicle; remove the moving and you have an electronics project. Both are worthwhile and neither is quite this.

That definition explains why the hobby demands three skills rather than one. Mechanics is structure, drive and the transmission of force, and it is where most first robots actually fail. Electronics is power, motor control and signals, and it is where most first robots fail expensively. Code is the part beginners expect to be hardest and it is usually not — the sensing and control loop is conceptually tidy, and the frustrating parts are the physical ones.

The honest thing to say about the subject is that it is very broad and the branches barely overlap in practice. Somebody building a competitive combat robot and somebody building an autonomous rover are solving unrelated problems with a shared vocabulary. Advice that does not say which branch it is about is usually the reason something did not work.

What makes it a robot

Sense, decide, move

Where first builds fail

Mechanics, not code

Cheapest real start

Microcontroller and two motors

The two real hazards

Solder fume and lithium cells

1. 1

   Step 1: Buy a documented kit rather than assembling a parts list

   A chassis kit with motors, driver, microcontroller and instructions is £100–£250 and removes the entire class of problem where nothing works and you cannot tell which of six unfamiliar things is wrong. The parts-list route is cheaper and is a much worse first month.

   You will know it worked when: You have a complete set of parts known to work together.

2. 2

   Step 2: Make one motor turn, and stop there

   Board, driver, motor, power. Nothing else connected. This is the smallest complete system in the hobby and getting it working teaches the thing that matters most — that power and logic are separate concerns.

   You will know it worked when: A motor turns on command and the board does not reset when it does.

3. 3

   Step 3: Add one sensor and close the loop

   Read it, print the value, and only then make the motor respond to it. Watching the number before trusting it is the habit that separates debugging from guessing.

   You will know it worked when: The robot changes behaviour because of something it sensed.

4. 4

   Step 4: Build a line follower

   The canonical first robot for good reason: it needs sensing, a control loop and drive, it is achievable in an evening, and it fails in instructive ways that are all about tuning rather than about mystery.

   You will know it worked when: It follows a line, badly, and you know why.

5. 5

   Step 5: Go to a club or an event before your second project

   Robotics has an unusually good in-person culture and an unusually large amount of tacit knowledge — what a motor should sound like, why a chassis flexes. A pit area at a competition is the fastest learning available in the hobby.

   You will know it worked when: You have seen other people's machines and asked how they did something.

• A microcontroller board — The brain, and the cheapest part of the whole thing. The ecosystem matters far more than the chip: choose one with documentation and a community, because your first twenty problems are all things somebody else has already solved and written up. Boards capable of far more than a first robot needs cost a few pounds.
• Motors, and the driver between them and the board — A microcontroller pin cannot drive a motor — it switches a signal, not power — and the motor driver is the component beginners omit and destroy a board learning about. Geared DC motors for drive, servos for jointed movement, steppers where position must be known without a sensor.
• A power source chosen for the motors — The commonest first mistake in the whole hobby is powering motors from the same supply as the logic and wondering why the board keeps resetting. Motors draw hard and sag the supply; the fix is separate supplies or proper decoupling, and it is a design decision rather than a bug.
• Sensors, starting with two — Reflectance sensors for line following, ultrasonic or time-of-flight for distance, an inertial measurement unit for orientation. Start with the minimum the behaviour needs — most beginner projects fail from too many inputs and no clear control loop, not too few.
• A temperature-controlled soldering iron and fume extraction — The iron is the tool that most repays spending on, because a cheap uncontrolled one makes joints that fail intermittently — which is the worst kind of fault to diagnose. Buy the extractor at the same time rather than later.
• A multimeter, and a bench supply eventually — The multimeter is how you find out what is actually happening rather than what you believe is happening, and it is the difference between debugging and guessing. A current-limited bench supply saves components by refusing to deliver enough power to destroy them.
• A 3D printer, sooner than you expect — Close to standard equipment once you want anything the catalogue does not sell — brackets, mounts, chassis parts. It is what turns "I could design that" into a part on the bench, and it changes the hobby more than any other single purchase.

And the ordinary workshop hazards, which are ordinary but not absent:

• Motors that are stronger than they look — A geared drive motor will take a finger into a belt or a wheel arch without noticing. Power down before working on a machine, and get into the habit early rather than after the first surprise.
• Bench power supplies and mains — Low-voltage DC is forgiving; the mains side of a supply is not, and neither is a large capacitor after the power is off.
• Lead solder and hand-to-mouth transfer — Lead-free is standard now and lead solder is still around. The exposure route that matters for hobbyists is hands to food, which is solved by washing them.
• Eyes — Snipped component legs travel at surprising speed, and anything with a spinning part deserves the same respect as any other workshop machine.
• Combat classes, which are a different category — Powered weapons are managed by event safety rules — arenas, link removal, activation procedures — and those rules are the authority. Nothing here substitutes for the ruleset of the event you are entering.

• Line following and maze solving — The classic entry point and a genuine competitive discipline at the top end. Cheap, tabletop, and unusually good at teaching control loops because the failure is always visible and always about tuning.
• Combat robotics — Machines built to disable each other in an arena, in weight classes from a few hundred grams up. Mechanically brutal, dominated by drive reliability and armour rather than by cleverness, and governed entirely by event safety rules. The most expensive branch, because it works by breaking things.
• Autonomous rovers and navigation — Machines that find their way without being driven, which is where the interesting software lives — sensor fusion, mapping, path planning. The branch with the most direct overlap with professional robotics.
• Walking and legged robots — Quadrupeds and hexapods. Mechanically demanding and servo-hungry, and the branch where a 3D printer stops being optional. Enormously satisfying and correspondingly fiddly.
• Robot arms and manipulation — Static rather than mobile, and about precision, repeatability and kinematics. Sits closest to the CNC and machining world, and appeals to people who want a machine that does a job rather than one that moves around.
• Educational and competition formats — School and youth leagues with fixed rules and standard kit. Worth knowing about even for adults, because the resources produced for them are among the best documentation in the hobby.

Robotics has an unusually good in-person culture for a subject that could plausibly have stayed online. Clubs, makerspaces and university groups host most of the activity, and competition pit areas function as open workshops where people will show you the inside of a machine they have spent a year on and explain why the second version is different. The tacit knowledge — what a motor should sound like, how much a chassis is allowed to flex — transfers almost entirely by being in the room.

The open-source norm is strong and load-bearing. A great deal of the hobby runs on freely published designs, libraries and build logs, and the expectation that you publish what you learned is genuine rather than decorative. It is also the reason the documentation quality for youth competition formats is so high, and why adults are well advised to read it.

Combat robotics is culturally somewhat separate, with its own events, safety officials and weight classes, and it is more of a spectator sport than the rest of the hobby. It is also where the engineering is most ruthlessly tested, because a design flaw is revealed by something hitting it.