Look to insects if you want to build tiny AI robots that are actually smart Roboticists could learn a thing or two from insects if they're looking to build tiny AI machines capable of moving, planning, and cooperating with one another. The six-legged creatures are the largest and most diverse multi-cellular organisms on Earth. They have evolved to live in all sorts of environments and exhibit …
Their order name, Ephemeroptera, comes from the seemingly ephemeral nature of their existence.
However, this does not take into account the fact that, prior to their emergence into the air, they have been living for months as water nyads (nymphs), gaily gamboling beneath the surface of streams and ponds, feeding, ever feeding, waiting for the day when they make their way through the surface film, wiggle free of their juvenile, aquatic bonds, gracefully take flight ... and finally get laid, lay their eggs and die.
AFAICS this isn't what the roboticists are interested in. They seem more interested in the apparently intelligent behaviour that arises from the combined activities of social insects. The units might be cheap but to achieve the results you'd need a lot of them, all doomed to end up as yet more electronic landfill.
Asimov got there first, a short story where USR addressed the growing public opposition to robots by developing minimalist insect robots that had such a small set of functions that the Three Laws could be trimmed down considerably, allowing for smaller and cheaper robots. Can't remember the name, and can't track it down.
Reminds me - anyone watch the Brazilian surveillance-society sci-fi series Omniscient? Weirdly stylish, no sign of a second series, unfortunately.
"The six-legged creatures are the largest and most diverse multi-cellular organisms on Earth."
If you're counting the entire colony as a single organism it has a lot more than six legs. If you're counting the individual colony members as organisms then the size is very strictly limited to the maximum that an insect's respiratory system can support.
Once I had a bumblebee nest in my wood shed. they would fly in and out through a roughly 50 mm knot hole at the bottom of the door. With the door wide open, bees would hover around inside and outside the shed, unable to find their way in or out. That's the result of early imprinting of a template which is subsequently rigorously applied to the 'decision' process. It's not comparable to even minimal human intelligence (which is adaptable to the unexpected) but does seem rather similar to the way much 'AI' works at present. Indeed a paper I picked up a while back suggested that bees (specifically) learn by drawing a tendency out of random choices on the basis of success rate - up to a certain point where the payoff is sufficient, and then the pattern gets locked down and behaviours ceases to be adaptable.
We should be glad that most humans don't operate on that basis.
I remember annoying almost everyone at an '80s AI & robotics conference by saying:
"when you can build an automaton that can do everything a bumble bee can do, to the same size and with the the same enrgy demands and operational range, you'll have accomplished something worth writing up".
They weren't happy with that.
1. All the robots want to be named Buzz, which leads to massive robot confusion.
2. The batteries keep running down when they do that little dance describing what direction and how far away the target is.
3. Insects are aggressive and territorial, so training the "AI" keeps resulting in aggressive and territorial robots.
4. Murder hornets hate robot insects just as much as they hate regular insects.
5. Paris wants to be the queen of the robot insects. ==>
Anyone coding in this environment, or a related one, needs to download and read the published paper, It's an excellent description and reading it will help everyone think about what they are doing and how it might (or might not) work. Understanding the environment and the way things work is a huge help when you are trying to get things working and coding. The paper is very helpful - that's a feature of so many published papers in every field you might have to work in... for example, this view is very helpful whether you like it or not (LOL): Mathematicians stand on each others' shoulders and computer scientists stand on each others' toes. - Richard Hamming
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2023
