What The Heck Is A Robot, Anyway?
In 1921, Karl Čapek wrote the play Rossum’s Universal Robots, thus coining the term “Robot” (ok, technically it was his brother Josef who coined the term ‘robot’, replacing Karl’s original suggested terms of laboři (from the Latin), or trudnik (from the Czech), but we won’t quibble over which one coined it. It was still Karl’s play. In the end, the root comes from the slavik for ‘worker.’) In the original play, robots were not electro-mechanical humans. They were very much flesh and blood like real humans, manufactured in fleshy parts and later assembled. This very much follows the golum and Frankenstein mythos. And it is clearly the basis for follow-ups like Do Androids Dream of Electric Sheep / Blade Runner, BattleStar Galactica, and to an extent, zombie mythos. Ah, but language is ever so fluid, and the original intended ‘Corpus Novum’ in the above tales has since been replaced by “clone” in modern usage. Yet we grandfather “robot” in on the above stories. But under Čapek’s original definition, none of us can call our creation a robot.
And now, we have so very many different opinions on what a robot is. Ask ten roboticists for a definition, and you’ll get fifteen answers.
- “It must move across the floor,” some will say.
- “It must have some grabbing device.”
- “It must have artificial intelligence,” other will say.
- “It must react to its environment or have sensors.”
- “It must do the job of a man.”
- “It must run without human intervention.”
- “It must obey the three laws.”
- “Must have an x-y-z table with a reprogrammable multipurpose manipulator”
- Etc.
Now, if I use any one of the definitions above (far removed from Čapek’s,) I can come up with something that fits the definition, but clearly is not a robot. For example, car welding bots are bolted to the floor. They have no lateral motion. Yet they are robots. But if I remote control such an arm, some would say it’s not a robot. At what point does a car with cruise control become a robot – when it can follow lines? When it can take you to work? Why doesn’t cruise control count as a robotic attachment? ‘cause GM’s marketing department chose not to call it that back in the 70’s?
“It must have a sensor and react to an environment” – well, every wall-thermostat has just that. And it’s reprogrammable. Analog? Sure, but it fits the definition. Everyone likes to call the mars rover a robot, yet people at JPL drive it. Hell of a lag time on those signals, but bundling packets of driving-coordinate data doesn’t make it autonomous. Many who call the rovers “robots” sneer at combat robots, yet they are fundamentally the same in operation as the Mars rovers. A Roomba is called a robot, but really, it’s mostly just touch sensors doing obstacle avoidance – how much intelligence is that? And why isn’t a dishwasher a robot if a Roomba is? “It has moving parts inside, and it’s reprogrammable, but no wheels” you say. Just like those car-welding robots w/o wheels that you do call a robot.
If I put a WiFi controller on a RoboNova, 99.9% of people will say, “that’s a robot!” Mostly because it’s in human form. Yet a lowly R/C (remote controlled) car with the same transmitter gets no such respect. Even when those R/C cars have sonar with ROV (remotely operated vehicle) override.
And let’s look at Asimov/Kubric’s HAL from 2001. No arms, no body. But he gets the tag. Those damnable bodyless web-crawlers qualify, but hexapods don’t? Millions of lines of code with not one motor, and both HAL and Google’s web-crawler lucks out for the title. C-3PO got into CMU’s robot hall of fame. I’m sure actor Tony Daniels – the human inside the C-3PO costume– was quite proud of being called a robot (although if the robot tag was to be given to a Star Wars actor, it clearly belonged to Hayden Christensen.)
For those few who’ve actually read Asimov’s short stories (don’t lie), recall that in one story, The Evitable Conflict, the robot has no body. It was just a machine. Some giant, pre-internet fantasy super-computer. Very clearly defined in the book as a robot. So to Asimov at least, a body was not necessary to gain the title – he used the term for both moving platforms as well as large-scale computers. Which makes al those damned spam bots grabbing email address of web pages “robots” by some people’s definition.
Eliza is often cited as a robot – but her AI (artificial intelligence) is so limited that she could not even gain an entrance exam into the Chinese box, much less pass it. Do we give AI’s an IQ test to see when they qualify for robot citizenship? Does a touch sensor make you a robot? A video camera? Does your microcontroller have to be above a certain processing speed to get you into the club? How many I/O (input/output) pins do you need to get an invite?
If I put a R/C PWM (pulse width modulation - a way of controlling motors by turning them on and off 60 times a second) controller on to two servos or ESC (electronic speed controller) controlled motors, and it moves by my joystick motions, many would call it a remote control car (or ROV if they’re being generous.) But everyone calls bomb-disposal bots or PackBots “robots,” even though they’re clearly the same basic R/C vehicles.
Add a speech recognition chip to one of the above – shifting from moving the joystick to the left to make the bot go left, to speaking the phrase “go left” – and now those same people will call it a robot. This is not autonomy. Does it really have any intelligence? Sure it has a sensor that listens for key words, but that’s not much difference from a PWM signal being received from a receiver. A 500 Hz voice signal=robot; but a 75 Mhz PWM signal doesn’t?
Full autonomy – lets return to the lowly household thermostat. It has sensors, it reacts to its environment, it has moving parts, and it even has electric and mechanical circuits. It controls hardware in your home with heating elements and blowers. But do we call it a robot? No. If we hadn’t had them until 4 years ago, I have no doubt that Helen Grainer would be marketing them as “Robot-Fireplaces.”
I think that we as robot builders have fallen into the trap that Supreme Court Justice Potter Stewart did in 1964 when he tried to explain what is obscene, by saying, “I shall not today attempt further to define the kinds of material I understand to be embraced . . . [b]ut I know it when I see it.” I think that this is where we as robot builders are clearly headed, which is a shame.
I’ve seen quite a bit of snobbery within all aspects of the field. Many Aibo programmers claim that they are the great roboticists because of the software they develop for the platform, and yet, many of those same programmers cannot replace a servo should it blow on the Aibo. This is akin to a NASCAR driver claiming to be a great mechanic, even though he’s never looked under the hood of his race car. Some fire-fighting bot builders sneer at combat builders, because the ComBot builders make ROV’s (yet most of those same people would call the Wood’s Hole’s AUV (autonomous underwater vehicle) “Alvin” a robot. (Is this an anti-wheel thing?)) Combat builders who design and build their own electronic speed controllers and drive-trains think of themselves as bot builders, but they sneer at anyone who has only assembled an off-the-shelf kit. Sumo robots come in both Autonomous and R/C classes. Is one class of sumo a “robot” while the other not?
So who’s right and who’s wrong? Where does the definition lie?
Well, like in programming robots, I don’t think it lies in a Boolean – “Is robot!” vs. “Is not robot!” Do we stop calling a paraplegic 'a man' because he cannot walk? Is a blind deaf woman no longer a person because she’s lost her sensors? Is someone who is brain dead no longer human, even though their heart still beats?
No boys and girls - a robot isn’t some narrow classification. I prefer to think of robots like "Americans." You don’t have to be tall or short, white or Asian, smart or dumb, strong or week, ambulatory or bed-ridden to be an American. And to be a robot, you merely need to have either some artificial intelligence (which is a whole ‘nother rant: does an SPST touch sensor really count as AI?), or some electro-mechanical body (do wind up all-wood automatons like the duck count? (I’d say so.)). Body and/or Mind. They are separable in humans and they should be equally so in robots. There will always be few people who get snobby about what’s what, even to the point of University vs. 'garage builders' classifying builders into 'who’s in' and 'who’s out'. But in the end, the line isn’t a line, it’s a long gradient of gray, slowly shifting from white to black, but heavily in the gray zone.
I think that we as a group need to move beyond exclusive feelings (you’re robot isn’t a real robot and mine is), and start with inclusive definitions. An analog is that engineering students at my University sneer down at the Business school, who in turn their noses up at the creative arts students, who don’t like to run with the kinesthetic majors, who in turn don’t like the geeks in engineering. But really, all these groups - they're all just people. They’re all students. They have more in common than not. And many bot builders (certainly not all) seem to moving down that path of separatism.
Can’t we all get along?
Robotics is a booming field for all of us. It is not a science unto itself. It is a nexus point of other sciences: Mechanical Engineering, Computer Engineering, Electrical Engineering, Computer Science, Kinesthetics, Chemical Engineering, and Art – to name but a few of the disciplines involved. In order for there to be great robots, we all need to work with each other, rather than locking yourselves into a little room one door down from some other guys who have already solved the electrical problem that you’re working on, while they are working to solve the mechanical one that you finished weeks ago.
I’m thrilled at the prospect of A123 lithium batteries hitting the street. Those guys are roboticists in my book – they’re making our bots more efficient with shorter recycle time and longer amp-hours. And batteries are to robots what hearts are to humans. We can’t have people without hearts. The art guys make robots more who are more emotionally available to the humans who will use them – if you don’t connect with the robot on an emotional level, will you really use it? Kinesiologists help me learn how to make my androids walk better. Look at what Bob Full at Berkeley has done to advance legged motion – and he’s trained as a biologist, not as an engineer. But he obviously qualifies for the title “roboticist.”
While, I’m not going to call a thermostat a robot anytime soon, I am going to continue to look at the field of robotics in terms of what cool things we can make together, and ignore the gangsta hand-sign that Tony Pratkanis throws the way of the ComBot arena. After all, someday his fire-fighting robot might need a more robust shell and ESC. And guess who’s got those?
Would a robot by any other name smell as metallic?
So let’s try to ignore what has become a much-misused term and focus on cool things we can make together (I’m sorry, but I do draw the line at EL-wire lined shirts being called “robot shirts.”)
Try augmenting your primary discipline with a secondary, and try working with other guys. One of the best teams I know of – Bob Allen and Ted Larson – come from opposite ends of the field: hardware and software. They fight like an old married couple, but the sum of their parts is clearly greater than the whole. And they’re better roboticists for it.
So I hope that will more people will start looking at other robots outside their normal comfort zone/specialty, and realize that they’ll be able to be a much better builder by studying cross-field. Michelangelo Buonarroti became a better sculpture because of his studies in oil painting. I bet you could learn something by doing the same type of cross-pollination within robotics (like shifting from microcontroller programming to welding 6061.)
Of course RoboGames is the best place to do that, but I’d never stoop so low as to put in a cheap plug like that, would I?






James (Sungjin) Kim
Invite as author
How about the origion of humanoid robot?