Rules for Roboticists

Rules for Roboticists


The following is an excerpt from my book, Absolute Beginner’s Guide to Building Robots (Que, 2004). I thought it might provide some helpful words of wisdom to anyone considering designing and building a robot. The illustrations here, a series of hero-engineer trading cards, were done by Make:’s very own Mark Fraunfelder and are peppered throughout the book. — Gareth

Rules for Roboticists

Remember The Rules from a few years back, that icky book filled with ridiculous, oppressive rules of dating etiquette? Well, I decided to dream up some rules of my own. No, they’re not things like “Never call a robot after the final assembly. Make it call you.” Or: “The way to a robot’s stomach is through its rear access panel.” These “rules” represent the collective working wisdom of builders who’ve been bolting together bots for decades. The cyberneticist Gregory Bateson used to say: “Always tie your ideas with slipknots.” So these are not hard and fast rules, more like rules of thumb. Just a few things to consider as you build your robots.


1. A roboticist is a generalist, a systems thinker.
One of the things that attracts a lot of people (me, for one) to robotics is that it involves the orchestration of many different disciplines. There are, obviously, specialists in the field — those who work only on AI control architectures, or robot locomotion, or whatever — but even they must keep the entire machine in mind. Most people who work in the field, and certainly all amateurs, have to have at least basic skills in numerous disciplines. As you get more into robotics, you’ll also find yourself spending a lot of time looking at humans and animals trying to figure out how they work. Oddly, trying to construct machine “creatures” gives one an even greater appreciation for the heavenly designs of nature, which brings us to…

2. A roboticist is a “deconstructionist”
As a robot builder, you’ll find yourself obsessively looking at the natural and built worlds and going: “Ah-ha! So that’s how it’s done.” Nothing will be safe as you take apart toys and machines that don’t work anymore (and some that still do), and find yourself playing with your food in a manner unsettling to others (“Cool, there’s the ligament attachments!”). But, for the love of all that’s civilized, leave the family pets alone!


Aside: Rumor has it that BEAM (and Wow Wee Toys) robot inventor Mark Tilden has been known to put all manner of bio-matter (chicken and other animal bones and bits) into his dishwasher so that he can clean them thoroughly for study of their mechanics, and one might even assume, incorporation into disturbing SRL-esque bio-mechs.
3. A roboticist knows how to K.I.S.S. it.
Actually not every robot builder knows this, but they should. K.I.S.S. stands for “Keep It Simple, Stupid” and it’s a maxim recited (but frequently unheeded) in many design disciplines. Heed it in your bot building. Take time to plan your projects. Don’t just throw technology at a problem ’cause you can. Use prototyping tech such as Lego Mindstorms, VEX, and breadboarding to test out your designs. Then try and figure out what you might not need and toss it out. The simpler and more elegant your designs, the more likely your robot is to be stable and robust.

4. A roboticist must learn to think “outside the bot.”
Innovation comes from thinking differently, heading down the road less traveled. Don’t be afraid to take chances, to go in radical directions. Apply what I call Rodney Brooks’ Research Heuristic. Here’s how this works: In his book Flesh and Machines, Brooks reveals how he came upon many of his radical ideas regarding robots and AI: He would figure out what was so obvious to all of the other researchers that it wasn’t even on their radar, and then he’d put it on his. Essentially, Brooks would look at how everyone else was tackling a given problem, and what assumptions were so implicit to them that these assumptions were no longer being questioned. And he would question them. Don’t listen when people tell you that you can’t do something. Ignore your critics.


5. A roboticist is as much an artist as a scientist.
Find someone who’s done anything truly cutting-edge in science and technology, and chances are, he or she has a bit of an artist’s/poet’s soul. Independent engineer and self-proclaimed “high-tech nomad” Steven Roberts is often quoted as saying, “Art without engineering is dreaming. Engineering without art is calculating.”


6. A roboticist must be methodical and patient (like any scientist).
The pressure that many robot developers are under to deliver creations that live up to sci-fi-like expectations leads too many to attempt too much, too soon. Scientific development is measured, by its nature. Don’t be afraid to get one thing right rather than a bunch of things “sorta okay.” (Notice how we just contradicted rule number 4. What can we say? Rules are … well for those two annoying ladies who wrote that book.)

7. A roboticist knows that neatness counts.
After you’ve built a few robots, you’ll quickly learn that the mechanics and (especially) the electronics can quickly become complicated, even in simple machines. There are usually wires sprouting everywhere, and trying to fit all of the parts inside your robot body, or on your robot platform, can become quite a challenge. You’ll learn that keeping everything neat and tidy will make a huge difference in the end. Use quick connectors when you can (for plugging and unplugging wires), use cable ties to bundle related wires together, and carefully plan (or revise) your design to maximize order and quick deconstruction/ reconstruction of subsystems for easier troubleshooting. Label. Color-code.


8. A roboticist must be a master of many trades.
As stated in rule number 1, a roboticist must be able to look at the big picture and know at least a little about a lot. He or she must have a working knowledge of materials sciences, structural and mechanical engineering, electrical engineering, and computer sciences. This may all sound intimidating to an absolute beginner, but knowing something about all of these areas of technology and science can actually be fun and exciting. And don’t let the big words trip you up. In plain English, these boil down to: building stuff (and knowing the right stuff to use), doing basic electronics, and knowing the ins and outs of microcontrollers and their software.

9. A roboticist should know his or her tools, materials, and processes.
You can have all the fancy “book learnin'” in the world, but if you don’t have a good working knowledge of robot building tools, building materials, and real-world construction techniques, you’re not going to be seeing robots scooting around your den anytime soon. The more you tinker, experiment, the more mad skills you’ll acquire — which leads us to…


10. A roboticist knows that you need to build early and build often.
Modern robot building technologies such as Lego Mindstorms, VEX, iRobot’s iCreate, open source microcontrollers, prototyping boards, and other similar innovations (not to mention computer designing, simulation, and programming software) allow robot builders a tremendous amount of freedom to experiment and build on demand. Think of pre-PC writing technology (pens and paper, typewriters) versus a word processor (complete with spell- and grammar-checking, a built-in dictionary, Thesaurus, and so forth) and that gives you some idea of today’s robot tools versus those of a decade ago, even five years ago. Now you can have an idea for a new drive or sensor system, whatever, and have it built and tested within a few hours. If it doesn’t work, you can quickly disassemble and assemble something else. From this rapid prototyping can come truly innovative robot designs.

11. A roboticist should know when to come back later (A.K.A. “The Kenny Rogers Rule”)
When you’re building anything, especially something as complicated as a robot, the build can sometimes get ugly. If you try to force your way through, you can often dig yourself into an even deeper hole. So here’s what you do: “Put the soldering iron down. Step away from the steaming robot entrails!” You’ll be amazed at what an hour away, vegging in front of the TV, rolling around on the floor with the cat, or sleeping on your problem will do. It almost never fails. Here’s a corollary: The extent to which you don’t want to drop what you’re doing and take a break (“I know I can fix this, damn it!”), is inversely proportional to the extent to which you need to take a break. Why is it the Kenny Rogers Rule? Cause as Kenny wisely tells us: “you got to know when to hold, know when to fold ’em, know when to walk away…”

[“Heroes of the Robolution” trading card illos by Mark Frauenfelder, from Absolute Beginner’s Guide to Building Robots]

14 thoughts on “Rules for Roboticists

  1. shing says:

    I am ashamed of myself for not owning this book. These cards are fantastic!

  2. auBois says:

    (a) The original title of the group was “Small Gods”, as taken from the book by Terry Pratchett. The generalized concept was that the original “old school” (ca. 1994) BEAM experimenters considered themselves “artificial life” creators and thus their devices “might” look toward them as “gods” of some sort. But there was NEVER a question of “God” vs. “god”. This IS an important distinction.

    (b) Tilden has NEVER produced a commercial product that actually used NV-net technology. He claims that the B.I.O. Bugs used a sound chip to “emulate” NV-nets, but he had previously discounted the possibility of ANY sort of digital processor being able to “properly” handle NV-net problems. That is an impossible dichotomy.

    (c) If you were to actually LOOK at Solarbotic’s catalog, you’d find an absolute paucity of BEAM-style kits. Oh, they now have lots of PIC-style ‘bots and lately they’ve gone gaga over the Arduino, but the actual number of Nv net kits you can build are now an exceedingly small list of formerly BEAM devices: This leaves you at approximately the 1997-8 level of sophistication. I think that Solarbotics is an absolutely great store (to whom I owe some grave apologies), but to consider it a “BEAM” store is somewhat questionable. Can you buy “Bicore-Experimenter” boards there any more? No. Can you buy the infamous and marvelous completely solar-powered “Pendulum of ‘DOOOOM’ there these days? Nope. There are quite a few others that are now missing, as well.

    (d) I think that BEAM concepts are still wonderful, particularly for people with small budgets. But as yet NO COMPLEX NETWORKS HAVE EVER BEEN BUILT USING THEM (Note to the hardcore types: ‘Hider’ isn’t really relevant and it is at least ten years in the past.) “Gadgets” that were looked upon as “complex” (a sphere using many tricores – “Möbius brain” as it was termed) had lots and lots of parts in it and spiffy LEDS everywhere, but ABSOLUTELY no scientific thought was performed with regard to
    “hypothesis, experiments, observations, nor even the glimmering of a theory”. Lots and lots of blinky lights but not a single word as to what it actually does – or more importantly – what could be done with it. No clue whatsoever as to what (or even WHERE) its input was, how that might be being processed, and what you could do with the output.

    About being nice: Well, there is “nice” and there is “absolutely journalistic”. You will find fairly few adjectives and adverbs in the above that might be construed as being un-nice: what you will find is merely a fairly objective journalist list of why Mr. Tilden’s opinions – which have varied widely over the years (anyone remember his joke about his ONLY use for a microprocessor as “being a comb for my beard?”) which provides no firm basis upon which to build a rule, a law… any firm foundation for thought.

    And, as a matter of fact, I -have- said these things to his face. Directly. He has grumbled afterward (or merely ignored me publicly), but he never appeared to have defended himself with regard to my statements. I rather doubt that if I was awful enough to him so as to so truly offensive that he’d wish to haul off and clobber me.

    The following is absolutely a truly STERLING value he possesses: that it IS true that he is a generous and an absolutely INCREDIBLY staunch and creative supporter of bringing robotics and technology to children, students, and other people who would previously never have imagined that they had the ability to create, with their own hands, imaginative and often useful or amusing devices: and even to do so on a truly limited budget. If you wish to consider THAT as a basis as developing a “law of robotics” (something like ‘ANYONE CAN DO IT IF THEY JUST TRY AND ARE GIVEN A LITTLE HELP’), then I’d support it utterly…

    1. Gareth Branwyn says:

      I am familiar with the origins of “Small Gods.” And I also know that Tilden has been, on occasions in the past, JOKINGLY referred to as the Big God of BEAM, as its inventor, hence the NICKNAME in these cards (which, by the way, are only half-serious, a fun way of presenting some ideas related to robots and the contents of my book).

      I appreciate your opinions on BEAM and Tilden. I too share your disappointment(?) that no complex systems have ever been constructed with BEAM tech (to my knowledge), but as you admit, it remains a wonderful little hobby, especially for those on a budget who want to build some fun little critters. And, I would argue that the sensibility of BEAM and bottom-up bot dev, has informed developments at Wow Wee, but YMMV.

      Re: Solarbotics URL on that card
      The book was written in 2003. At that time, Solarbotics almost exclusively sold BEAM components. I understand they’ve expanded their offerings greatly today. And Tilden at the time basically had no website, so Solarbotics was the main game in town.

  3. auBois says:

    Stated: “The book was written in 2003. At that time, Solarbotics almost exclusively sold BEAM components.” Uh, I hate to tell you, but NOBODY SELLS BEAM COMPONENTS AS SUCH, SPECIFICALLY DESIGNED FOR THAT PURPOSE. FOR THAT MATTER, IF BEAM WASN’T EVEN MUCH OF AN ESTABLISHED FIELD “WAY BACK WHEN”, THIS STRENGTHENS THIS ARGUMENT FURTHER. Those companies might sell motors, they sell batteries, they sell solder, they sell ICs, they sell transistors, they sell diodes, they sell perf-board or other electrical construction board, they sell LEDs, they sell solar cells, they sell IR transmitters and receivers, they sell hobby-related wheels, etc., ad nauseam. They did not at that time nor do they now sell BEAM-exclusive (or, for that matter, even BEAM-more-or-less-related) parts. Even the GM-xx motors and stuff came originally from Wow-Wee and Wow-Wee has (I repeat) never created a BEAM-based robot (i.e. runs on ANALOG Nv/Nu-net technology: no “digital black blob” included).

    Solarbotics merely went to the catalogs, started buying whatever they thought they needed with whatever money they had and started selling it. If people wanted to build BEAM stuff with it: fine, they could. If people did NOT want to build BEAM stuff with it: likewise, fine.

    However, “almost exclusively sold BEAM components” is nonsensical. Even if Solarbotics had not existed (a very sad thought indeed), one COULD HAVE BUILT PRECISELY THE SAME DEVICES AS IF THEY DID.

    Praytell, do you agree with this or not?

    Oh – about Wow Wee: RS Media runs off of Linux (it was so touted, anyhow). I simply cannot see that as being “BEAM” in any way whatsoever.

    Also: Tilden never had a website. There was a horrible website at LANL, but it was created by someone other than Tilden.
    Tilden has an ____expired____ patent – expired in both the United States and Canada. For some reason that I simply cannot fathom, his most fervent acolytes continue to stick in warnings that you have to get his permission before using the technology, which, of course, is not true.
    Tilden’s papers, few as they are, are not in the least bit substantive: I’ve got them all, and the only useful material in ANY of them come from the people who worked for him. There is nothing in them that he contributed that has any real substance.
    Mark Dalton created the first BEAM mailing list; when he had to stop, I took it up and then other hands took it over in April 2000. All told, since about 1976 THE COMMUNITY [which essentially does not include Tilden] has written far in excess of 73,926 articles (read it, ya’ll – Seventy Three Thousand, Nine Hundred and Twenty Six And Then Some Articles), of which Tilden contributed perhaps 20..40. It is reported (I’ve read articles in which he has written…) that he was not “willing” to put any “articles” on the list because either (a) he didn’t have the time to write them or (b) the community would not understand them.

    When I finally was able to leak the schematic for the BEAMAnt that was used in a project at a yearly technical conference in Colorado, not only were the electrical characteristics of the entire device analyzed within one week, but also a rather detailed listing of the behaviors the device would exhibit given various bias-resistor set-ups and “environmental” factors. The concept that a community that could write as many articles as it has done, and a community that has shown some unbelievably bright, hard working people who not only (a) analyze anything they can get their hands on, (b) invent, build, document, and describe anything new they can possibly imagine, (c) help and answer any question that a “newbie” might write and (d) ACTUALLY HAVE REAL JOBS OUTSIDE OF THIS and just keep doing this year after year during which time the so-called “Big BEAM God” has contributed essentially nothing really calls into question the so-called nickname.

    I termed myself “The Official BEAM Heretic” years upon years ago. You no doubt can detect my stridency here. Part of it comes from the fact that Mr. Tilden has promised “his people” much and has delivered nearly nothing. Much more, I’m concerned about people who somehow are swayed by the nickname and thus believe anything that Mark Tilden says.

    For example:

    He has told many variations of his “White Sands Mine Detection and Destroying Robot” story.

    One is that there was one and only one built. However, supposedly a test – including destruction – occurred. Oddly, a robot of precisely the design given hung up on the Los Alamos National Laboratory Museum wall for 5..7 years. Somehow I find a discrepancy here.

    Secondly: the robot had a nearly crippling design flaw. Let us ignore, for the moment, that most mines these days are made of plastics and have essentially no ferrous materials in them. Tilden’s “Mine Stomper” had what was, in effect, two long antennae, each with a device that was -extremely- sensitive to magnetic flux variations. Well, you can already see that would be a bit of a design flaw: how can it “see” mines for which its detectors are not designed. FAR worse than this, however, is that the electrical motors that drove the legs produced SO MUCH MAGNETIC NOISE that you couldn’t take a reading at the same time as you were moving. At best, the robot could move “forward” (it could be in a forward curve depending on if the flux detectors actually picked up a differential), THEN its motors could be turned off, and THEN it could take a reading and determine what to do next. To put it another way: we are describing an incredibly slow robot, cooking in the broiling sunlight of White Sands, with quite limited battery power… my guess is that it would run perhaps 12 hours before the energy was gone.

    Third: Mr. Tilden has come up with this horrifically melodramatic story of late. Supposedly some officer, a Colonel (or other ranking officer) I believe, was on hand to watch the test. Let us remember that in “The Real Military” that a Colonel might have to send in a bunch of HUMAN BEINGS into an area where they are likely to be killed and maimed: really, really nasty stuff. Supposedly… the Stomper went in and stepped (somehow luckily) on a mine, setting it off. “Some” still effective portion of the robot continued struggling along until that part set off another mine. At this point, there was the “poor little mechanical beast” just trying to move a little bit further on in order to continue its mission. At this point – and get this, folks – supposedly the Colonel just couldn’t take it anymore. He had been so emotionally affected that he called the test off – he couldn’t bear to watch it continue. Forgive me, friends and neighbors, but I come from a military family. No such bleeding heart would ever be put in charge of personnel who might be injured or destroyed. Period. Oh yeah – have any of you watched the footage on YouTube that shows things being blown up in Iraq? Y’know, there is nothing quite like seeing a center-of-the-road bomb being able to flip an American tank completely over, utterly shearing the turret off. That is a couple of tons of metal isn’t it? The energy to cause that kind of damage is unbelievable. And yet, Tilden expects us to believe that his little ten-pound robot would be able to continue to be in one coherent chunk after such an explosion… and even continue to function. Finally, remember that the Army pulled funding from Tilden’s project. For a while (and I heard this with my own ears), his explanation was that the soldiers didn’t want to have this job taken away from them. This is odd – they say that sappers are truly mad in a fairly serious way: they are willing to perform a job where they can be blown to bits JUST BECAUSE THEY ARE PERFORMING THE JOB. At least a “normal” soldier who has been properly trained and is in a group of other such solders don’t expect such a thing. So tell me, how could sappers… even as insane as they must be… want to continue to perform a job that some machine could do? I simply am not willing to believe that.

    1. Carnes says:

      I don’t know that much about tilden but mines/bombs, yes. Anti-personel mines can be plastic (but metal is likely for old mines) and anti-vehicle are metal. An anti-vehicle mine is looking for significant weight or magnetic field (an above surface tilt-rod can also be used), not a little robot. A burried anti-personel mine won’t take much weight and if there are above-surface elements to the mine a piece of trash caught in the wind could set it off.

      It’s possible a robot could survive an anti-personel mine (which are designed to injure, not kill). If the mine was designed to “bounce” out of the ground before going off it’s possible the robot could survive. But most landmines are laid far enough apart to not set each other off. Any damage to the robot that significantly impared mobility would probably keep it from reaching another mine. A normal in-ground blast mine would probably completely destroy the robot or throw him a good distance away if he had armor of some sort.

      From your story, i doubt the testing involved live mines. Blowing up prototypes is a waste of resources. A live test would be performed at some point, but not on the only existing robot and probably not with an entire field (mines are expensive too).

      But, i will say that it isn’t silly that the robot had to move so slowly. Mine detection is slow and tedious for a reason. Nobody will spend the time to do it anyways.. located a mine? blow it in place. Located a field? mark it and go around. Can’t go around? use the MICLIC

  4. Robot Platform says:

    I promise I will try to follow rules… lol…
    Just liked this statement “Rules are … well for those two annoying ladies who wrote that book.”

  5. Robot Platform says:

    I promise I will try to follow rules… lol…
    Just liked this statement “Rules are … well for those two annoying ladies who wrote that book.”

Comments are closed.

Discuss this article with the rest of the community on our Discord server!

Gareth Branwyn is a freelance writer and the former Editorial Director of Maker Media. He is the author or editor of over a dozen books on technology, DIY, and geek culture. He is currently a contributor to Boing Boing, Wink Books, and Wink Fun. His free weekly-ish maker tips newsletter can be found at

View more articles by Gareth Branwyn
Maker Faire Bay Area 2023 - Mare Island, CA

Escape to an island of imagination + innovation as Maker Faire Bay Area returns for its 15th iteration!

Buy Tickets today! SAVE 15% and lock-in your preferred date(s).