Tiny Robots At Play, In Words And Pictures

justin sane writes: "The NY Times has an article about the one of the tiniest functioning robots to date.[Note: free reg. req. ^[?] -- t.] They faced numerous problems and build the robots layer by layer with photolithography on expoxy compounds. The microprocessor is raw (i.e. without a package to save on size). The batteries are the biggest part by far (not surprisingly). There is an MPEG of one in action as it's speed 20 in/min velocity but alas it just looked like a photo album on my M$ player--still the photos are cool. No word if they are working on a port of Embedded Linux that can run the 8k memory space though. That would be my next step, then ... Python ;-)"

Comments

No reg. req.

[Anonymous Coward]

The old "partners" trick stopped working, but you can read the article by replacing the "www" in the URL with "channel".

which would lead to

[hawk]

tthe death by a trillion pin pricks . . .

But they might be quite useful if we're ever invaded by armed paramecia . . .

Even better, we could use them for mounts for the Yeast Brigade, so it would be cavalry . . .

;)

BEAM robots...

[JawzX]

Some would argue that BEAM type analog robots (invented by Mark Tilden) are not really "fully functioning" but they do exhibit quite a deal of complex behavior and even a rudimentry short term memory due to feed back loops in their simple Nv/Nu transistor networks. Well any-way the point is that thanks to SMT packages and the simplicity of the circuits involved these autonomous little electronic creatures have been nearly this small for years! oh yeah, and usualy solar powered... here's a couple of pictures of robots ive seen...these are all simple photo-tropic ostacle-avoiding devices with tactile and IR sensors... http://www.beam-online.com/Robots/Galleria/Large/o pticPopperFL.JPG http://www.beam-online.com/Robots/Galleria/Large/T hermoWpenL.JPG http://www.geocities.com/frankendaddy/Smiley.html http://www.sierranet.net/~polygon/twitch.html these aren't even the smallest i've seen, just some of the better sites with pictures. if you want more infor on how to build your own BEAM type robots, www.beam-online.com is a good place to start looking.

Re:You just know...

[HamNRye]

Hmmmm....

Does Battlebots have a super-lightweight division???

Please...

[Uberminky]

First of all, didn't we just see this recently? I could be wrong, maybe it was some other news source.. Too lazy to go search for it.

But come on folks. Did you look at the thing? It's really not all that impressive. Sure, it's kinda cool. But impressive in any way? Not really. They mention all the nice techniques they use to miniaturize the electronics, and yet after all that work it has nothing but one single temperature sensor?

I am working on an extremely simple, very tiny robot myself. Very similar to this one in fact. I have access to none of the expensive things these guys do. In fact it will probably cost me about $20 total. Granted, it will probably be slightly larger. But it will also do more.

I don't mean to be so negative.. And in fact my response isn't so much to the article or the guys doing the work on the robot, rather than to the comments I've seen on Slashdot. Come on, folks. A machine that can barely climb over a dime on a level table is going to have a bit of trouble traversing even the smoothest parts of Mars. Even using them for surveilance would be silly. (Although this is actually one thing I'm thinking of playing with, simply for fun. Have it seek out audio in the range of human speech, and relay back what it hears. <evil laugh>)

The hardware is really not all that impressive. I have beside me two pager motors, which I'm using as the main drive system of my own tiny robot. They are roughly half an inch long, and about as thick as a pencil. Now align them in parallel, one with the shaft facing left and the other with its shaft facing right. Attach them to some small "tank treads" and you're in business. A drive system in about 1/2" square. The device they show doesn't even seem to need a microcontroller, it could be built to do the same with analog electronics. My mini robot has a "brain" of an 8-pin microcontroller, which has 6 I/O pins, leaving me plenty of room for expansion (light sensors, bump sensors..). Certainly it will get larger than the device shown. But not by very much, and its functionality would be superior (in this incarnation, at least). Sure, they have plenty of work left to do. I just think that we're all getting a little too ecstatic about something that's very very simple...

Shameless plug: Go to Indiana University? Join the IU Robotics Club! We're just getting started up.

Coins

[morie]

There we are, cruisin' along, and all of the sudden people start putting coins in our way! Luckily, being a miniature monster truck, we can take those babies!

seriously, it looks very cool, but will it actualy *do* something? Will it cary tiny loads etc?

Re:Huh?

[OlympicSponsor]

Okay, less glibly, consider how you and two friends would move a sofa if there were some compelling reason not to communicate. One of you goes to one end of the sofa, the other two might abort a move to that end, one gets the other end, and whoever didn't make it to an end of the sofa opens doors and whatnot. It's really not neccesary to say "I'll get this end" except to be social.

But this only works as long as we both have the same plan in mind. It's obvious that both ends of the sofa have to be lifted and that the door needs to be opened. But what happens when we get to the door and the sofa needs to be turned? I start turning it on it's back so you follow suit, great. Still doesn't fit. I have the idea standing it on end, so I put my end down. Thinking I want to rest, so do you. Now I have to come over to your end to lift it up--but I can't, the sofa is in the way. I lift my end up instead--and you lift yours. OK, we'll just move the sofa back in the room and turn around, that way I (the only one with the right plan in mind) will be on the right side of the doorway. We do that. Now I need to lift my end up to stand it on end, but you need to put yours down--how do I make you do that?

With robots, I suppose the second actor could match the first actors actions against some kind of internal list of strategies to see which one it is employing. Then it could take the role labelled "helper". The trouble is, flexibility requires more scripts which entails more possibility for misunderstanding on the part of the second actor. It would take forever to get anything done. Simply put, acting in concert requires an overarching plan. If that plan is non-obvious, communication is required.

And here's the ultimate argument against communication-free robots: They don't exist in nature. If evolution could have gotten by without building the incredibly complexity of the human language system by some means as described above, it surely would have. Ants communicate via pheromone trails. Bees communicate with dancing. If they could have done without these systems, why don't they?
--
Non-meta-modded "Overrated" mods are killing Slashdot

Re:Why batteries over other energy sources?

[OlympicSponsor]

Electromagnetic induction power transmission will probably interfere with radio communications (between units, from control to units and in the surrounding area). But you've got the right idea. Why not use solar? This only works in the sunlight (duh), but if we are talking about exploring other planets (as another poster mentioned) that's not such a big deal.

Of course, if you want to implant these somewhere (a body, a mine, etc) that won't work so well. OTOH, neither with broadcast power, especially underground.
--
Non-meta-modded "Overrated" mods are killing Slashdot

Yuck, I'd hate...

[OlympicSponsor]

...to be on the receiving end of THAT scheme.

There I am, standing around outside. *clunk* An interplanetary space probe lands next to me. *sssss...chink!* It pops open. Out "spews" (your word, not mine) a horde of tiny beetle-sized creatures. Creepy.

In any case, they aren't necessarily all that great for "exploring new habitats". Think about it: they are too small to move anything or contain any internal testing equipment. You probably can't put a lens on the thing that's big enough to show you more than about 5 grains of sand. It might be OK as a low-cost first wave to determine average temperature, etc--but you really need a rover-style device to actually EXPLORE.

However your "collaboration" comment gives me an idea. Shoot a bunch of these guys at the planet, say 1 million of them. In parallel, they determine the 100 most promising areas of the planet to concentrate on. The split into 100 teams of 10,000 each and home in on the centers of those areas. Only about 1 in 10 will make it and those 1000 join together (panda-monium!) to make a much larger unit capable of doing more extensive work.
--
Non-meta-modded "Overrated" mods are killing Slashdot

Huh?

[OlympicSponsor]

"Essentially, the idea was that robots could use much the same logic they use to navigate an arbitrary set of obstacles to co-operate to acheive a task without needing to communicate anything."

Ummmm...could you expand on that? I don't see how "logic used to navigate obstacles" is communication.
--
Non-meta-modded "Overrated" mods are killing Slashdot

Re:Huh?

[OlympicSponsor]

"Have you ever really watched a load of ants try to move a leaf ? If you watch closely, you will see them pushing the leaf every which way and from every direction. A simple hypotheses is that any movement problem is easier to solve some ways than others and eventually the ants with the right idea 'outvote' the ones with the wrong idea."

And in the end, what did they accomplish? They moved a leaf. Big deal.

Off the top of my head, I can think of three ways to run a "many agent" system:

1) Like ants as you describe. The problem is that if the only workable strategy requires going over a "hump" in parameter space, this won't work. Think of it like evolution. In order to reach a goal, every mutation has to make sense on the way there. Evolution can't do something now to "lay the groundwork" for something later. Same with ants--if they need to perform some non-obvious preparation step it just won't happen. Adding intelligence could help with the foresight issue, but intelligence isn't enough to solve this problem because there's no way to coordinate plans so everyone knows WHAT preparation needs to be done.

2) Hardwired strategies. But we all know how inflexible those are.

3) Communication with at least one intelligent agent in the group. Now an over-arching plan with foresight can be created which will then be executed with the rest of the group essentially being extensions of the intelligent member's body. With humans, this is exactly what happens. I say "put your end down and hold it while I lift mine". Or you think of something and say "turn it over backwards".

What I should have said in my previous post was "communication-free robots that perform difficult, non-obvious tasks don't exist in nature".
--
Non-meta-modded "Overrated" mods are killing Slashdot

Re:Don't limit yourself to "nature"

[OlympicSponsor]

"...CPUs, RAM, DVDs, TCP/IP...don't exist in nature."

Don't they? Brains, memory, language all fullfill the same function. "Robots in nature" ARE fantastic at communication...we're doing it right now.
--
Non-meta-modded "Overrated" mods are killing Slashdot

In other words,

[hawk]

You want to build a Beowolf with these things . . .

:)

hawk

Can anybody tell me...

[joshv]

why this things needs a microprocessor? All it seems to do is move forward, not much bit munching needed for that. It did not even appear to do anything even as simple as tracking a light source.

Maybe they are running some sophisticated AI code like the following:

while (battery_PCT()>10) {
move_forward();
}

-josh

Re:Collaboration

[commanderfoxtrot]

Surely the aim of this research is to see how small it is possible to go and then make it larger?

Now that it has been shown how to make such a very small robot, the techiniques &. learnt can be used to make more practical robots the next size up, with more power, commmunications and logic systems.

Can you imagine...

[Goldberg's Pants]

The show "Robot Wars" using these? They'd have to put it on Pay Per View, the matches would last hours.:)

---

Re:Hardening and radio capabilities, then... space

[andr0meda]

Frankly I don`t think putting things like this on the Marsian soil is going to give us a propper mapping of the surface. The terrain they can cover is practically unsignifficant, I suspect they`re not really protected from crashing into holes or gliding from slopes, and their broadcasting resources are limitted. It`s probably better if we could put bigger and more robust wheeled vehicles on the surface since they have better navigation capacities. But I agree that the distributed peer 2 peer broadcasting approach would be a nice angle :)

Why batteries over other energy sources?

[A1kmm]

15-20 minutes is probably not enough time for the robot to last in the tasks advocated in the article for it to complete, so why not use some sort of alternate energy source - perhaps electromagnetic induction from an "energy transmitter"? It only needs a relatively small energy intake, and it will only move relatively small distances from where it is deployed.

Hmm, this seems familiar

[s00bf153]

Anyone remember Inner Space. God, this thing could cure my bad back.

Forth Is Still Good For Something???

[Baldrson]

Amazing that Forth might still be good for something. I didn't think they made memories that small anymore. I mean 8K bytes is about what Western Electric was limiting the Viewtron videotex terminal to in 1981 when I tried to get it's ROM burned with Forth so we could dynamically download graphics macros. Efficient use of RAM as well as 1200bps modem bandwidth. (This is, in fact, what gave rise to Postscript subsequent to our visit to Xerox PARC.)

Collaboration

[Nyarly]

When I was in school, one of my profs had done his dissertation on the idea of robotic collaboration without communication. Essentially, the idea was that robots could use much the same logic they use to navigate an arbitrary set of obstacles to co-operate to acheive a task without needing to communicate anything. Prof's name was James Jennings, and he pretty much made my upper level education worthwhile.

It seems that this sort of behavior would be ideal for these little bastards. I mean, they're cute and all, but radio transmission would zap their batteries, and coordinating communication would eat all their memory. But a small growth in memeory size would be enough to manage the sort of navigation algorithms I recall.

Unfortuantely, mass produced these things are supposed to run a couple hundred USD, each. Tough to rationalize many handy applications for that. Even cat entertainment seems to be a little overkilled by that sort of investment (although they do seem ideal, don't they? Maybe not fast enough.)

Ushers will eat latecomers.

You just know...

[MWoody]

Somewhere, in a bunker deep within Area 51, a team of scientists is already working on a tiny little gun to stick on this thing. Ah, well, it's a sure way to get funding.
---

Hardening and radio capabilities, then... space

[MikeCamel]

We finally seem to be getting down to a decent size for planet exploration. Harden them up a bit, and drop them over a terrain. They shouldn't need much cushioning, as they'll weigh very little.

Give them a single burst radio broadcasting capability, so they can report back, and you're away. Cheap, light, low volume - how many can you fit in a Mars probe? This should be great way of exploring new habitats.

One alternative to just dropping them is to land, and then spew them out - that way you get a lot more detailed information about a small area, and can control them so that if one of them finds something interesting, they can all go and investigate. Next step, of course, is to allow them to collaborate, and decide to do it themselves.