Digit works in a warehouse and moves boxes from a cart to a conveyor belt. He has legs that look like they came out of a Dalí sketch, but they work. He lifts up to sixteen kilos, and for a good part of the time he does it without saying a word (so to speak). Then, every now and then, he decides that gravity is an opinion, and he lies down on the floor like an exhausted puppet. It also happened at a fair: impeccable work, attentive audience, and then… free fall with the box flying away. Sipario. As long as it stays behind a laser barrier, everything is fine. But the goal is to get these humanoid robots out of there: into offices, hallways, shared spaces. And there, if it gets its balance wrong, it's not enough to raise an eyebrow: you need rules. Real ones.
The red button that doesn't help
In classic robots, there's always him, the big red button. Press it, and time stops. But with humanoids, things get complicated: turn off the power and he wilts like a plant forgotten on vacation. Only he weighs sixty-five kilos and has less flexible joints.
Agility Robotics, which makes Digit, is looking for an alternative to the sharp blow: gentle shutdown sequences, where the robot slows down, puts things down, kneels down and lies down with a certain dignity. A bit like us when we realize that they're about to ask us a technical question and we have no idea of the answer.
Humanoid robots yes, but without masks
One of the key issues is to understand thing is meant by “humanoid robot”. Does it need a T-shaped body? Or is it enough that it can walk upright and interact with us without knocking us down?
The IEEE proposes to forget about aesthetics and think in terms of behavior. A robot is humanoid if it lives in our world: if it walks like us, it moves where we move, and risks stepping on real feet. That's why we need new rules. Not those of robotic arms on assembly lines, but standards that adapt to robots that could follow you down the corridor without warning.
Boston Dynamics, Agility and others agree: We set the outcome, not the path. If the goal is to do no harm, the including It's the designer's business. As long as it works.
Humanoid robots, speak clearly. Or rather: signal well
If a robot decides to cross your path, it can't do it silently. It has to say so. Not with theatrical phrases, but with clear signals. Digit's little lights are a good start, but in a noisy warehouse the effect is that of left-on indicators: everyone sees them, no one understands them.
You need a visual grammar for robotic movements: legible postures, announced changes of direction, predictable logic. Like the indicators of a car or the passive-aggressive tone of a colleague on Monday: you immediately understand where he wants to go.
Voice might help, but if there are eight robots in the room all talking at once, the result is not collaboration. It's algorithmic karaoke.
The risk is all in the face
When a robot has two (fake) eyes, a (stylised) mouth and maybe a (default) smile, we tend to think that sit. And if he ignores us, we get angry. But the problem is not only emotional. It is that we treat him like a person, we let our guard down, we are too close to him. Some of us he's even falling in love.
And if in the meantime he is still on page one of the manual "how not to drop things", it could end badly. The IEEE, for this reason, also proposes a reflection on aesthetics: the robot must to appear for what it is. No more buddy masks if all he can say is “Error. Try again.”
Better rules today than accidents tomorrow
The point of all this is not to block innovation, but to give it a framework. A standard that says: this is the minimum. Below, there is no going down. Above, everyone does the best they can.
Today, humanoid robots are still learning to walk. Literally. But between warehouses, hospitals and public spaces, the step towards coexistence is short. And if we want to do it without stumbling, we should write the rules now.
As one of the ISO managers says, that is working on a new standard, if in the end all the producers are equally dissatisfied, maybe we have found the right standard.
Besides, better a few complaints now, than a robot asking you “Is everything okay?” while you pick up the pieces of the broken tablet.
