SteelCollarRobotics and Aeronautics
Ground JoeAir AmeliaOrbit KatherineGovernments Abe
Home / Essays / What Asimov got wrong about the factory floor

Plain talk, with jokes

What Asimov got wrong about the factory floor

· 4 min read

The Three Laws of Robotics are eighty years old, beautifully written, and useless for buying a palletizer. The real first law of a working cell is shorter and less poetic: a certified integrator signs the risk assessment, and nobody gets to skip that paragraph.

The short answer

Isaac Asimov published the Three Laws in 1942, in a short story called Runaround, and they have shaped how people imagine robots ever since: a machine may not harm a human, must obey, must protect itself. They are wonderful fiction. They are also a description of a robot that thinks about you, and the robot in your plant does not think about you at all. It knows where its joints are, how hard it is pushing, and whether a safety circuit is closed. Everything else is the integrator's job, and the standards that govern that job are written in the dullest prose in the English language on purpose.

Where the Three Laws actually live

Not in the robot. In the application. ISO 10218 and the American R15.06 standard cover stationary arms. ISO/TS 15066 covers collaborative applications, the ones where people and the arm share space. R15.08 covers industrial mobile robots, which have their own ways of surprising you. None of these documents ask the robot to care. They ask a competent person to look at the cell, the part, the gripper, the people and the floor, decide what could hurt someone, and engineer it out or guard against it. That person signs. That signature is the first law.

The practical consequence for a buyer is simple and a little deflating. A robot marked collaborative is not a robot that is safe. It is a robot with features that make a safe collaborative application possible, if the gripper, the part, the speed and the layout allow it. A collaborative arm holding a sharp casting at speed is not a collaborative application, no matter what the brochure says, and the assessment will say so.

The second law, obedience, is also not what you think

Asimov's second law has robots obeying people. Real arms obey programs, and programs obey whoever wrote them, which in most small shops is the machinist who got sent to the two day class. That is good news. It means the cell does what your people tell it, in your shop, with your parts, and it will keep doing it at four in the afternoon when a person would start cutting corners. The bad news is the same sentence read the other way: it will also keep doing the wrong thing, precisely, until someone notices. The robot never has a hunch.

The third law, self preservation, is mostly a maintenance schedule

A robot protects itself with limit switches, collision detection and a service interval. It does not flinch. If the gripper is wrong for the part and the arm is driving into a fixture every cycle, it will report a fault, stop, and wait for you to fix the fixture. Which, to be fair, is more self control than most of us show.

What the movies cost you

The expensive mistake is not believing in killer robots. Almost nobody does. The expensive mistake is the opposite one: assuming that because the arm is friendly, round edged and marked collaborative, the safety work is done. HAL 9000 had a pleasant voice too. The shops that get hurt are the ones that treated the risk assessment as paperwork rather than engineering, usually because the seller let them.

What to do with this

When you talk to anyone about a robot, ask who will perform the risk assessment, which standard it will be done under, and whose signature will be on it. If the answer is the robot's manufacturer, the dealer, or you, keep looking. The right answer is a certified integrator who has built your application before. On this site that is the rule the concierge follows every time, and it is why Joe will tell you about fit, payload and payback all day and will never once tell you a cell is safe.

Describe your station to Joe if you want to see that in action. He will give you the honest read on whether the job is automatable, and then introduce the humans whose names go on the paperwork.

Related on this site

Fiction gave robots a conscience. Industry gave them a risk assessment, a safety-rated stop and someone whose name is on the paperwork. Buy accordingly.

Put it to the test on your own job

Tell Joe what you are running and he will apply this to your numbers, including telling you when a robot is the wrong answer.

More notes