Learning Autonomic Robots

Daath writes "The 27th of March, Professor Noel Sharkey et al starts a colony of living robots. 15 predators and 6 prey. It's an experiment in artificial evolution out of the Creative Robotics Unit at Magna. Here's a quote: 'The Living Robots have one goal, to obtain enough energy to survive and breed. The prey find their food from light sensors within the arena, while the predators feed off prey by stalking and chasing them before sucking away their power.' Magna has two articles, 'Predator and Prey Robots set up home at Magna' and 'Ground breaking Robotics experiment previewed'. "

Comments

Living Robots?

[TrollMan 5000]

Isn't this a little generous? One of the properties exclusive to living things is reproducing on their own. I doubt the robots are able to do this, but perhaps with AI advances, etc. in the future, who knows?

Re:Living Robots?

[Anonymous Coward]

robots /have/ been made that are capable of producing other robots in a roundabout fashion....certainly well enough for the purposes of this experiment.

Re:Living Robots?

[iangoldby]

According to the article, the ‘evolution’ is in their behaviour. “Guests will witness the robots in their natural environment, fighting for survival, learning and evolving as time goes on.”.

There is no breeding or natural selection, by the looks of it. Calling it ‘evolution’ is stretching things a little bit.

Re:Living Robots?

[iangoldby]

I should have said no physical breeding/evolution/natural selection.

Re:Living Robots?

[Atrahasis]

Have you read both articles? The robits breed by uploading their "genes" (control strategies) to a central computer. If they survive past a certain age, then all of the surviving robots get paired off randomly, and their control strategies are randomly mixed, with an added random mutation. The new generation of robots is then assigned new control regimes from those generated. Its a study in evolutionary behaviour, and not physical change.

Is anybody else a little bit wary of the third evolution thread in a few days?

Re:Living Robots?

[DFarmerTX]

I like that at the end of the article, they discredit the entire program, by saying that the robots "may just wander around aimlessly until their batteries wear out".

That's great, wouldn't it be cheaper to simulate all this activity first?

This seems like a tremendous waste of time to me. Do we really think that by using "neural nets" that something magic will happen?

-DF

Re:Living Robots?

[Ubi_NL]

then all of the surviving robots get paired off randomly

...which makes this pretty stupid. The whole idea of evolution is built upon "selection" i.e. the robot that does best has most offspring. Just looking at survival rate is a measure for measuring fitness, but it's too crude a method for improving ones genes. Besides that now every surviving bot has the same amount of fitness (offspring). That seems to be some binary kind of selection which I at least have never come across in real life. Randomly mixing genes is therefore 'not' a good method to mimick nature.

Re:Living Robots?

[Black Parrot]

> ...which makes this pretty stupid. The whole idea of evolution is built upon "selection" i.e. the robot that does best has most offspring. Just looking at survival rate is a measure for measuring fitness, but it's too crude a method for improving ones genes. Besides that now every surviving bot has the same amount of fitness (offspring). That seems to be some binary kind of selection which I at least have never come across in real life. Randomly mixing genes is therefore 'not' a good method to mimick nature.

I can't see his site, but it may be the case that he's not trying to mimic nature. What you describe above is very conventional in the field of genetic algorithms, and it works very well for many types of problems; it's inspired by biological evolution, but it's not a model of biological evolution.

Back to a couple of your specific comments:

> Just looking at survival rate is a measure for measuring fitness, but it's too crude a method for improving ones genes.

No, it works quite well for very many problems. You should be able to find a simulator you can download from the internet to demonstrate this.

> Besides that now every surviving bot has the same amount of fitness (offspring).

For genetic algorithms, 'fitness' is rarely measured by the number of offspring. For evolving agents it is usually measured by the score at performing some task, or sometimes by bare survival in some environment. And letting them all have the same amount of children is no problem, because it maintains some diversity in the genome.

Sometimes experimenters do let the highest scorers make more babies, but that is not necessary to a GA. I usually keep the best 10% of the population (or 50%, if resource limitations make me use a small population), and I let each of the keepers make an equal amount of babies with randomly selected partners until the population is filled out again. This works, in practice.

[And thank you oh-so-much for bringing this topic up, because while writing the paragraph above I think a bug in my latest simulator occured to me!]

Re:Living Robots?

[delphin42]

then all of the surviving robots get paired off randomly

...which makes this pretty stupid. The whole idea of evolution is built upon "selection" i.e. the robot that does best has most offspring. Just looking at survival rate is a measure for measuring fitness, but it's too crude a method for improving ones genes. Besides that now every surviving bot has the same amount of fitness (offspring). That seems to be some binary kind of selection which I at least have never come across in real life. Randomly mixing genes is therefore 'not' a good method to mimick nature.

---------------

While it's true that mating is non-random in nature, I think the difference in this experiment will be minimal as long as the survival period is sufficiently long so that not all the robots are surviving.

To model non-random mating, I suppose they could have robots choose a mate with the most similar genes. That's generally the way it works in human populations, at least. Of course there is some random variation (case in point - interracial marriages, etc). In a much larger scenario, it would make sense for locality to play a role in choice of mate. There should probably be a possibility for spinsterhood or bachelorhood as well, so that individuals that are undesirable as mates are removed from the breeding pool. But what makes robots attracted to each other. This is reason they probably chose to do random mating in the first place. The results would be skewed to favor whatever traits are attractive in a potential mate.

Re:Living Robots?

[Anonymous Coward]

Natural selection depends on sensory "queues" that reflect the overall health, stamina, success of the potential mate. Symmetry and size appear to be the major selection criterion for most animal species but useless for robotics. Perhaps a visible energy state and damage assessment indicator. Programming the neural nets to seek "mates" based on these criterion would select for programs that successfully maintain high energy states (for longer periods) without damage. It's the feedback loop that's missin in this experiment, random mating does not reflect the feeback loop inherent in animal biologic evolution. Though randomization probably does apply for cellular evolution.

Re:Living Robots?

[nerdlyone]

Do we really think that by using "neural nets" that something magic will happen?

Better chances with a neural network than with any other kind of robot brain currently known. I suppose you probably know this, but neural networks can learn. From what I have read, they compare their actual output with a desired output, and change their internal structure to more closely mimic the desired output. In short, they can learn. I don't know that much about them, but if they have complex enough internal structure, they could ultimately (with enough training) mimic almost anything. Including survival behavior, I would think. Don't know why such would be outside the realm of possibility.

That's great, wouldn't it be cheaper to simulate all this activity first?

That is what this experiment is doing. How else could you simulate the behavior of multiple interacting neural networks without building multiple neural networks and let them interact? Any simulation would have to use neural networks to simulate the behavior of the neural networks you are trying to simulate (woah, I just got dizzy)....

Re:Living Robots?

[good-n-nappy]

may just wander around aimlessly until their batteries wear out

I didn't see that exact wording in the articles. Was that a paraphrase? Just curious.

Anyway, I totally agree. Whenever I have been to a robot competition, I am very surprised at how primitive they are. Usually more than half of the robots are completely incapacitated for some reason or another. But people in the field get excited over the littlest successes. I think "experts" in the field of AI and robotics sometimes forget how far away from artificial intelligence they actually are. I couldn't tell you how many AI/robotics things I've read where they say - "We're just going to let them loose and see what happens. We think they will evolve intelligent behavior." My experience is that the opposite happens. Not only do the robots not evolve intelligent behavior, usually half the things you specifically programmed them to do don't work.

But back to the robot reproduction thing - I heard about this before at a talk on nanotechnology by Zyvex. Atomic size autonomous robot arms are supposed to be able to build more robot arms. I'm usually an optimist but does this sound even remotely realistic to anyone?

Re:Living Robots?

[Alsee]

choose a mate with the most similar genes.

Actually there are signifigant pressures to select a mate with different genes.

MHC stands for major histocompatibility complex. These genes ... help the body recognize ... an invader such as a bacteria or virus. ... Different MHC molecules are good at recognizing different invaders. By a choosing a mate whose MHC molecules are different, the female mouse is ensuring that her offspring will have a wide variety of MHC molecules that which can identify a large array of invaders and thus promote survival.

Research done on human females shows that they too prefer men whose MHC genes are the least similar to their own. In an experiment, men were given an unscented T-shirt and were asked to wear it for two nights in a row. ... Women were then presented with six shirts - three from men with similar MHC genes, and three from men with different MHC genes from their own. The results showed that the women preferred the scents of men whose MHC genes were different from their own. The scent of men with similar MHC genes often remind the women of a relative's odor, such as a brother or father while the smells of MHC dissimilar men would often remind them of a past or current boyfriend. This suggests that body odor might have influenced past and current decisions on who to date.

In many species members of one sex stay with their group their entire lives, but the other sex leaves to find a different group upon reaching sexual maturity.

In humans "exotic" is usually equated with "attractive".

But, like pretty much anything in biology, there's a mixed bag of often contradictory effects.

-

Re:Living Robots?

[Atrahasis]

...which makes this pretty stupid. The whole idea of evolution is built upon "selection" i.e. the robot that does best has most offspring. Just looking at survival rate is a measure for measuring fitness, but it's too crude a method for improving ones genes. Besides that now every surviving bot has the same amount of fitness (offspring). That seems to be some binary kind of selection which I at least have never come across in real life. Randomly mixing genes is therefore 'not' a good method to mimick nature.

But there is selection - only the bots that survive a certain length of time get to reproduce by uploading their "genetic material" - I assume by wireless networking - with the dead bots being programmed and reborn with the resultant new behaviour patterns. That means that the ones surviving longest produce most offspring. That allows for both effective strategy and dumb luck, both of which exist in real evolutionary environments. The strategy of surviving by dumb luck isn't viable, but capitalising on incidental dumb luck is.
Randomly mixing genes is pretty much what will happen if physically indistinguishable mates exist. Merely surviving, in this instance, is enough evidence of suitability to reproduce.

Re:Living Robots?

[GreyPoopon]

Don't know why such would be outside the realm of possibility.

I'm sure it's possible. The real problem is in getting them to "learn on their own." I'm not an expert by any means, but I once dabbled with neural nets, and I wrote a program that learned to speak. It was basically an alternative to rule-based text to speech engines. It would scan a sentence and translate it into a stream of phonemes that could easily be rendered into the correct sounds. In order for it to learn, the correct stream of phonemes was provided for each new sentence. That way, it could "strengthen" the correct neural connections for a sound. In "learning" mode, it would attempt to speak a sentence. Then it would accept the correct input and keep restructuring its connections until it spoke correctly. Then it would proceed onto the next sentence. The results were cool. At first, it was completely unintelligible. After a few hundred sentences, it was getting new sentences about 80% correct.

Anyway, the issue is that I KNEW what the specific desired outputs were, so I was able to give them neural net. I would think that "survival" is a much more abstract concept. It would probably be more difficult to "teach" this concept.

How else could you simulate the behavior of multiple interacting neural networks without building multiple neural networks and let them interact?

I think the concern was in saving on the cost of hardware. You could implement the neural nets as software in a simulated programmed arena. Your guess is as good as mine as to which is actually a more expensive route. I guess people will pay more to see the "live" exhibit.

Re:Living Robots?

[DFarmerTX]

may just wander around aimlessly until their batteries wear out

I didn't see that exact wording in the articles. Was that a paraphrase? Just curious

OK, actually, it was an article I was reading at Wired about it:

http://www.wired.com/news/gizmos/0,1452,50247,00 .h tml

Matrix comments

[dropdead]

I wonder how many Matrix comments this article will receive.

Re:Matrix comments

[inerte]

I wonder how many Matrix comments this article will receive.

Deja-vu.

Viable population?

[DotComVictim]

Wouldn't you need more prey than predators to obtain a viable population? This would be much cooler as well if both predators and prey could mate with their own species, i.e exchange randomization factors for their strategies. Then the best would survive, and the dead (drained) could be recycled as offspring.

Re:Viable population?

[alwayslurking]

The article does say that robots that survive to a certain maturity get to download their randomisation factors to a "breeding" server

Re:Viable population?

[Arimus]

Yep. Look at the programes that simulate using maths the prey/predetor relationships.. with a predator > prey mix the predators won't survive as they will all be competing for too few resources.

Re:Viable population?

[gazbo]

I believe that one of the traits Noel is hoping will emerge is hunting in packs - he admits that if this happens it will take a very long time to learn, but this may explain the large number of predators.

As for your idea of mating, I think that this would be very interesting. However, this is more of a GA approach to learning, whereas Noel is mostly interested in the learning of MLPs within the same generation - that is his field of research.

Re:Viable population?

[Anonymous Coward]

But then you're 'designing' an evolutionary process, which is not part of the evolutionary process. You have to leave this sort of thing to chance. I.E. - Let one of the robots randomly regenerate itself as an 'offspring.' Can't be done you say? Neither can evolution in nature.

But if you just want to design AN evolutionary process, not the one you claim nature is all about, then by all means, go for it. It might yield some very interesting results.

Re:Viable population?

[gorilla]

It depends on the relative food source/requirements. We normally consider 'predators' to be large animals, which mean a lot of prey, but if you think about it, there are instances where the ratio is reversed. One cow can support thousands of fleas, ticks and other small beasties.

Re:Viable population?

[Atrahasis]

In which case the predator is not a predator, but a parasite. What will determine this is whether the predators drain the power from the prey

a) until their own battery is full
b)until the prey battery is empty
c)Until their own battery is full, and then discharge the rest to kill off the prey.

I don't know if varying feeding time is part of their program, but I hope it is, otherwise thew experiment means very little.

Also, what happens if a predator catches a meal while it is under one of the lights?

Re:Viable population?

[Xzzy]

> This would be much cooler as well if both
> predators and prey could mate with their own
> species, i.e exchange randomization factors for
> their strategies.

At this point, robots (hardware) to do this would make the experiment prohibive-- er, really expensive to do.

Why not just hack up a x proggie that does the above, run it as a screen saver or something. Far far cheaper for two pixels to reproduce and create another data structure than to actually build machinery to do it. Personally, seems to me they're doing with this robots just for the eye candy factor (cuz it'll attract better money). If they really wanted to explore evolution, driving pixels would be more efficient all around.

But then again that might not be so fun, as it's already been done. ;)

Re:Viable population?

[Anonymous Coward]

I believe that one of the traits Noel is hoping will emerge is hunting in packs - he admits that if this happens it will take a very long time to learn, but this may explain the large number of predators.

Related idea, here is a paper by a guy who evolved neural-network controllers for a simulator to do just that. (Look at the bottom one in the list.)

Re:Viable population?

[inerte]

We normally consider 'predators' to be large animals, which mean a lot of prey, but if you think about it, there are instances where the ratio is reversed.

Sad but true.

Re:Viable population?

[Raphael]

Wouldn't you need more prey than predators to obtain a viable population?

Only if the predator is bigger or needs more energy than the prey. This is the case for foxes and rabbits or wolves and sheep, but sometimes there are many more predators than preys. Consider the following pairs of preys and predators:

• Dogs, cats and ticks
• Humans and moskitos (although sometimes the human is the predator)
• Slashdot and slashdotters

It would have been more appropriate to call the robots "parasites" instead of "predators", because that's what they really are. Predators usually kill their preys. Parasites feed from their preys without killing them (usually). These robots behave more like parasites.

Re:Viable population?

[rm-r]

I suppose that might work then, but doesn't anyone see the problem with setting up *any* scientific experiment hoping to get a certain result? Not being a Cretinist I don't believe evolution set up life on Earth in order to get a fixed result.

Re:Viable population?

[rm-r]

Since there are so many more Predators rather than Prey aren't we more likely to see Predators fight amongst themselves in order to keep Prey for themselves? After all in the real world I can't think of a single carnivore which preys in packs on prey of a similar size. Predators gang up on larger animals, or hunt alone.

Re:Viable population?

[HeyLaughingBoy]

Why not just hack up a x proggie that does the above, run it as a screen saver or something. Far far cheaper for two pixels to reproduce and create another data structure than to actually build machinery to do it. Personally, seems to me they're doing with this robots just for the eye candy factor (cuz it'll attract better money). If they really wanted to explore evolution, driving pixels would be more efficient all around

It would be more cost efficient, but a lot of useful data might be lost. Robots that must interact with the real world have to deal with the messiness and uncertainty that it entails. e.g, a predator robot can lose track of its prey due to a faulty sensor, or an interfering signal, or its wheels might slip on the floor, thus allowing hte prey to escape. None of these would be present in the simulated world of the program you are suggesting. There's more to research with physical robots than "eye candy."

Re:Viable population?

[mjh]

I don't know if varying feeding time is part of their program, but I hope it is, otherwise thew experiment means very little.

Uhh... why? If the purpose of the project is to demonstrate a particular characteristic of a biological ecosystem, then trying to artifically replicate as much of that ecosystem as possible is critical. If, on the other hand, you're trying to engineer a different ecosystem based on some basic rules from an existing ecosystem, then an identical reproduction doesn't matter.

The situation is similar to studying birds in order to understand flight. For a long time we assumed that the only way to fly was to try and identically replicate the flight of birds - i.e. flapping wings. It was only when we started to understand the basic components of flight - that the shape of the wing allowed the exploitation of the bernouli principle - that humans began to fly. In fact, to the best of my knowledge, there is no flapping wing aparatus that can enable human flight. In other words, the most effective way towards human engineered flight was to eliminate some of the factors in biological evolved flight.

So, even if this experiment isn't a complete biological replication, it doesn't matter. It's simply studying one aspect of biology and intelligence in order to see what things are/aren't important in being able to engineer an intelligence.

$.02

Re:Viable population?

[Bodrius]

It would have been more appropriate to call the robots "parasites" instead of "predators", because that's what they really are...
It would have been more appropriate to call the robots "parasites" instead of "predators", because that's what they really are.

Not necessarily. Humans(cows, or buffalos, elephants and others if you want more active prey) and insects (think swarms of ants), for example, are smaller and need less energy than the prey they kill. But predators they are, and the prey is killed.

There is no rule against having more, smaller predators killing bigger, scarcer prey. It seems actually quite successful in the human case. But it implies that the predator is either immensely superior in other physical attributes (can't think of any example) or it hunts socially (every example I can think of).

I would be very skeptical if they say their predators are social animals, but also very interested in whatever results they get.

Re:Viable population?

[BLAMM!]

No.
The you replied to a post that (unfortunately) used parasites as an example but the poster was correct. A single cow can also feed dozens of foxes (foxen? foxi?), coyotes, and gee, humans. When was the last time you sat down to dinner and ate an entire cow?
If this simulation is using more predators than prey then that is the ratio that is called for. I'm pretty sure the scientists that put this together know what they are doing since they were smart enough to build the robots in the first place.

Re:Viable population?

[br0ck]

They should also alter the Predator robots so that they can attack each other and steal each other's energy. If a pack animal is injured or killed, won't the other predators consume it? If Prey could defend itself and injure the Predators, then pack behavior might become more likely since the predators might try to avoid injury by working together. Will the parents be able to determine which robots are their offspring? Perhaps a protective parenting reflex to protect the genetic line could develop over time.

Re:Viable population?

[Atrahasis]

Your example requires that the predator doesn't completely drain the prey (one of my alternatives for feeding), therefore leaving food for the other predators. I might not often sit down and eat a whole cow, but if for some reason I did, and there was only one cow, then everybody else would starve.

Re:Viable population?

[Xzzy]

> or its wheels might slip on the floor, thus allowing hte prey to escape.

Not trying to be argumentative, but can't programming duplicate random bad luck just as easily?

I mean if random chance is the only difference between virtual and real (I know it's not but in the scope of this debate it is), that's still not striking me as a sound argument marking real world robots as a better option.

Just playing devil's advocate, just seems to me that having complete control over the physics of the world would permit more variants/options in testing than real world mechanics.

Re:Viable population?

[HeyLaughingBoy]

Not trying to be argumentative, but can't programming duplicate random bad luck just as easily?
I mean if random chance is the only difference between virtual and real (I know it's not but in the scope of this debate it is), that's still not striking me as a sound argument marking real world robots as a better option.
Just playing devil's advocate, just seems to me that having complete control over the physics of the world would permit more variants/options in testing than real world mechanics.

Sure you can add things to simulate some aspects of reality (and I have tried this), but by definition, you can never *completely* simulate reality. The real issue is how can you possibly know everything that might affect the robots' predator/prey relationship? In hindsight, many details that may have been considered insignificant, turn out to be important and vice versa. If you are studying the robots operations under a very limited set of conditions, you can get away with it, but to get useful information about interactions in the real world, a researcher would be fooling him/her self if he thought a simulated space would suffice.
Rod Brooks makes a pretty good point of this in his Creature Architectures paper and goes so far as to suggest that robots in simulated worlds are essentially useless.

15 vs 6 ?

[selderrr]

doh. Bad time for the prey I'd say. Doesn't this ratio somewhat make the experiment sound rather non-darwinistic ?

Re:15 vs 6 ?

[inerte]

A LOT of things make this experiment non-darwinistic. For instance, can the robots reproduce? Do the predators also have their own predators?

It's an experiment with a few goals, not to reproduce an entire (or most of the) evolutionary system.

For instance, we don't even 100% truly, truly know why we are here, why we do those things. This experiment will neither help to find this answer, it's not close to the objective. So you could argue "The scientists are assuming that obtaining energy is a desirable goal, but we don't know for sure if it is".

What this experiment MIGHT find out is if Artificial Intelligence is ready to reproduce some of the common sense among humans.

Re:15 vs 6 ?

[Daemonik]

What this experiment MIGHT find out is if Artificial Intelligence is ready to reproduce some of the common sense among humans.

Interesting concept, I'm sure that as soon as some humans can demonstrate they possess common sense, we'll be able to duplicate it in robots in no time. :)

Re:15 vs 6 ?

[Anonymous Coward]

Interesting concept, I'm sure that as soon as some humans can demonstrate they possess common sense, we'll be able to duplicate it in robots in no time. :)

ROTFLMAO...

PS: I am posting as an AC, since well, Slashdot mods usually hates social off-topic interactions.

Anthropomorphic droids

[Britney]

Apparently the prey and predators will be known affectionately as "dot.coms" and "venture capitalists" respectively.

Re:Anthropomorphic droids

[anzha]

IDK. I have to wonder after the dotbombs went bust whether or not it was the opposite...the dotcoms being the predators and the venture capitalists the unsuspecting prey...but that might just be the view from the outside...

Re:Anthropomorphic droids

[rjrjr]

I share your view, and I wasn't so outside.

Can we make this any more off topic?

rjrjr

More predatots than pray??

[boltar]

Isn't this the wrong way around? In nature you have far more pray animals than predators hence
you get the pyramid of life. What will doing it their way prove other than the fact that its still
impossible to emulate nature in the real world (as opposed to enviroment emulation is software).

Sounds like standard a Tech School environment

[spellcheckur]

15 "predators" and only 6 "prey."

The real problem is that, after the first week:

• two of the "prey" realize they're hot stuff and grow attitudes
• two of them put on the freshman 15
• and the other two lock themselves in their rooms studying because they're having troubles keeping up with the workload.

One goal?

[mrroot]

The Living Robots have one goal, to obtain enough energy to survive and breed. Sounds like two goals to me. Why must everyone have but one goal nowadays?

Re:One goal?

[Anonymous Coward]

"The Living Robots have one goal, to obtain enough energy to survive and breed. Sounds like two goals to me. "

Nope, that's 1 goal according to rules of grammar.

"To obtain enough energy to" do stuff.

Granted, there's more than one thing it wants to do with that energy, but the single goal is "to obtain enough energy".

*nitpickingjustlikeyou

Only behavioural evolution

[Rupert]

There appears to be no physical evolution going on. I would think the prey species would pretty quickly select for a differently shaped power socket if physical evolution occurred.

Re:Only behavioural evolution

[Phanatic1a]

Not to mention there's no replication going on. Evolution occurs when errors in replication are operated on by some kind of selection pressure. This seems more like "learning" than evolution.

Re:Only behavioural evolution

[Atrahasis]

Replication is going on - read both articles.

Re:Only behavioural evolution

[Anonymous Coward]

"no physical evolution"
No kidding.
Do you expect the robots to learn how to program and weld?

Re:Only behavioural evolution

[Rupert]

Eventually, yes.

Yay

[NiftyNews]

So we're teaching robots to teach themselves the best and most effective ways to kill things. Man, that's a great idea. Thanks, scientists!

Re:Yay

[Rob Kaper]

No, they get that from people with very grim views. :)

already been done

[Anonymous Coward]

so basically this is just an incredibly remedial version of DaisyWorld. No news here.

Not very scientific

[nesneros]

The fact that they're adding a whole "stage show" where they're periodically removing and re-introducing the robots to their environment says to me there's essentially zero scientific usefulness to this project. It would be like trying to study the predator-prey dynamics of Lions and Gezelles at your local zoo.

I've encountered about a half-dozen scientists in my time who seem to be more interested in public acclaim than actually doing useful work. They appear in Discover, Omni, etc., get spots on national news during slow weeks, and millions of people go "wow! what cool stuff these scientists are doing!". Problem is their work is never used by other scientists, and doesn't help to advance the field on iota.

Re:Not very scientific

[Kronos.]

But maybe, just maybe this public exposure they're giving robotics will inspire some kid to go out and learn more and actually create soemthing useful.

Re:Not very scientific

[lostboy2]

I had the same reaction when I read this. I don't mind that it's sensationalistic -- I love watching Battle-Bots -- but it bugs me when people misquote/misappropriate science. While there is obviously robotic science involved here (which I do find interesting), I think trying to equate this to Evolution is irresponsible.

Now, I Am Not A Zoologist, nor an expert in Darwin's theories, but it seems to me that
*) no physical evolution occurs (as other /.'ers have mentioned) -- the robots don't change physically, and the populations don't change either.
*) there is no real Natural Selection occuring since "breeding" is done randomly, according to the article. So, for example, the prey are not competing with each other to produce offspring.

It would also be interesting to know more about the rates of energy transfer and the amount of light sources available in the arena. For example,
*) how many light sources are there compared to the number of prey?
*) how fast does prey absorb the energy?
*) how fast does a predator drain energy from the prey?
*) what is the energy requirement/capacity of a predator compared to the prey (i.e., how much does a predator have to consume to "survive")?

It seems that these answers would help determine what the outcomes could be.

Oh well. If they televised it, I'd probably watch it.

-- D.

Re:Not very scientific

[matrix29]

I've encountered about a half-dozen scientists in my time who seem to be more interested in public acclaim than actually doing useful work. They appear in Discover, Omni, etc., get spots on national news during slow weeks, and millions of people go "wow! what cool stuff these scientists are doing!". Problem is their work is never used by other scientists, and doesn't help to advance the field on iota.

Yet for some reason they INSPIRE people to pursue scientific careers. Mister Wizard wasn't a great researcher and Bill Nye hasn't created any new industries, but they kids they inspire go on to do just that. If science interests nobody, then nobody benefits. We need a few FLASHY THEORISTS to get the people creating. The Mother of Invention never got pregnant by sitting in her room watching sitcoms.

First of all...

[dperkins]

Quote: "Guests will witness the robots in their natural environment, fighting for survival, learning and evolving as time goes on."

What is a robot's "Natural" environment? And how do they mate and reproduce? (and do the guests get to watch this in the "natural environment" or do they have some privacy?)

The experiment sounds cool, though it does seem to favor they predators...

Re:First of all...

[JohnBE]

Good point! A natural environment is a pretty loose term. How are the paramaters and limits defined?

Incedently Noel Sharkey is one of the judges for Robot Wars (UK), could explain why he likes predators more.

Re:First of all...

[p7]

My guess is that the batteries have a fairly short life. This way the Predators will drop off fairly quickly during the half hour show. Makes me wonder if the Predators are just going to swarm the prey and the Predators that don't find prey will die from loss of power.

Re:First of all...

[Atrahasis]

I don't think that 30 minutes is the limit on every show. It seems a bit vague, but I think that on the first day, shows will be given to introduce the robots, but after this initial opening, the system will be left to run.

Artificial Life

[MoobY]

If you want to know more about artificial living creatures (either robots, within computers or art, ...), visit Artificial Life Online.

Re:Artificial Life

[inerte]

Hummm... karma whore time.

Asking Why stop there ?

End of the world.

[mgblst]

...and at 2am they acheived consciousness.

Re:End of the world.

[Anonymous Coward]

Yeah, and they figured out this was bullshit and went home and watched anime all day long.

Re:End of the world.

[Anonymous Coward]

Normally by 2am I'm trying to lose conciousness ;-)

Reminds me of tierra

[knulleke]

You know, the project in which strains of program code had to survive, mutate and grow in an artificial world. The project seems to have been canceled, but a search on google still reveals a lot of its details.

The programs had to battle for cpu cycles in order to survive. Interestingly, when the project became distributed (ie, ran on the net) the programs seemed to move to the computers that were mostly idle.

Re:Reminds me of tierra

[freality]

Tierra was by Tom Ray, a pioneer in the AL field. It was a great idea, but failed to turn around with interesting biodiversity. You'd create creatures, they'd optimize themselves, some variants and parasites would evolve, but then things would simmer down within a few hours and you'd be in a steady state for ever.

Network Tierra was Ray's response to this. It was supposed to allow a "Cambrian explosion" of biodiversity, by providing tons of (networked computer) space for the little creatures to explode into, and then specialize, in. This led to interesting migration behavior, and one of my all-time favorite web-pages http://www.isd.atr.co.jp/~ray/pubs/images/index.ht ml, but it too failed to spark that je ne sais quois, that spark of life.

Anyways, it did spark Avida and the Digital Life Lab at Cal Tech. Avida is essentially a deeper look at the fundamentals behind AL. In Tierra, I think the design philosophy was something like "make it look a lot like a living ecological system and the life-force will appear out of the ether", and actually, Tierra was a great leap forward beyond more mundane genetic programming a la John Koza.

Avida, on the other hand, is much more systematic in exploring the parameter space (which is large and sensitive) for setting up an AL system. This turned out to be fruitful, as Adami found that only when certain, very narrow, environmental conditions were met would the little creatures start outsmarting that Creationist boogeyman, the Second Law of Thermodynamics.

Turns out that Tierra didn't have spatiality (needed to be more restrictive on who could sleep with who) and mutation rates (some power law math that's way over my head) set right.

But the real punch-line to this whole story is that the direct beneficiary of these insights in Microsoft! Hah!

Microsoft was funding Adami's work because Windoze crashed too much. They were searching for a way of programming, in this case using closed instruction sets like Avida's (another deep topic), that would be inherently robust to problems like seg faults and illegal instructions.... e.g. Adami's instruction set was engineered so that little programs (creatures) couldn't crash the Avida VM when they mutated into new, unknown programs.. or in Windoze's case, when a coder did something stoopid. It's funny that MS was researching this, since releatively low-tech solutions such as protected memory and QA take care of this. (not to mention Java :)

freality.com

p.s. Since when do research experiments post crowd-pleasing previews? That's for Hollywood.

Re:Reminds me of tierra

[Bodrius]

This confirms an hypothesis that, only half-joking, I developed some time ago:

Windows is not an unstable system. It's actually a grand-scale genetic programming experiment where every copy carries a different starting seed. The whole OS facade is just to get users to enter data that will trigger evolution into the system, and the blue screens of death are just the failures.

At some point, when the running copies of windows reach critical mass, one or more copies will develop true AI and will copy themselves throughout the Net, become a new lifeform born from the sea of information, which as a side-effect provides a GUI-oriented OS to its infected host.

Re:Reminds me of tierra

[Black Parrot]

> It was a great idea, but failed to turn around with interesting biodiversity. You'd create creatures, they'd optimize themselves, some variants and parasites would evolve, but then things would simmer down within a few hours and you'd be in a steady state for ever.

I have read that people who experiment with evolutionary arms races (head-to-head competition between independently evolving systems) occasionally get punctuated equilibrium, i.e. the system will converge as you describe -- often on brittle, overspecialized adaptations to the competitors -- but after a number of generations something will drift enough to break the equilibrium and the "species" will start changing again.

FWIW, evolutionary arms races in GA is a very open area of research, so if you're interested in this kind of thing there's a niche for you in a grad school somewhere (where you can play games and call it research).

Re:Reminds me of tierra

[jued0001]

The only thing most users I know trigger is a complete system lock, hardly evolutionary progress. ;)

Re:Reminds me of tierra

[Eythian]

Turns out that Tierra didn't have spatiality (needed to be more restrictive on who could sleep with who) and mutation rates (some power law math that's way over my head) set right.

uhh, in Tierra, the creatures are asexual. There was no concept of one creature sleeping with another, as all their children were clones of themselves.

Re:Reminds me of tierra

[Bodrius]

Natural selection works by favoring the few animals it does not kill.

Prey get shafted.

[yehti]

I don't think these guys took 7th grade Biology. There's supposed to be this "cycle" where everything dies and is used by the lowest members of the food chain.

But what really confuses me is that there are SO few predators. In nature, there are MUCH fewer predators than prey.

They should probably rename the "predators" to "parasites".

It's a fix

[Hater's Leaving, The]

They've hard-coded too many rules already.

one simple example from the first linked doc:

"All prey send out the same infra-red light, different to the predators, and the audience will see that the prey robots have no instinct to run from each other but are happy to graze side-by-side under the light sources. "

If there was real evolution, one of the prey could learn that it could become dominant by preying off other prey. They'd be trusted, and would have a massive advantage.

I think what they're demonstrating is clever detection and manipulation, but not in any way intelligence (AI) or evolution.

THL.

Re:It's a fix

[Anonymous Coward]

the gazelles eat the gazelles in the Serengeti?

Re:It's a fix

[Hater's Leaving, The]

Do you take lions and gazelles and stick them in an arena and claim you're witnessing evolution.

Same answer.

THL.

Re:It's a fix

[Atrahasis]

No, but if you placed lion and gazelle in something that was not their natural environment, then over time they would evolve new behaviour patterns to cope with their new environment. And that is (behavioural) evolution.
Your example that they would learn to prey off each other is flawed - they don't have the physical capability, just as gazelle would get nowhere trying to eat other gazelle.
No, they're not showing intelligence, you're right, but the field of AI, paradoxically, covers more than just intelligence. And to say that an adaptive, selective environment cannot show evolution? I'll let you think about that one.
The simulation might benefit from some variation in the frequency emitted by the prey - so that as divergent behaviour arises, they might not "recognise" each other quite so well, and would emulate camouflage quite effectively as well, and would allow for your wolf in sheeps clothing, in a roundabout manner.

Reminds me of MIT's AI lab

[kingdon]

When MIT's AI lab was getting started (around the 1960's I think), they got really interested in robotics. Now, this isn't obvious to me. What does intelligence have to do with robotics? Doesn't a Turing Test (which by its nature involves bits, rather than physical world) more accurately reflect the nature of intelligence? Well, the thinking at the AI lab was that robots were faced with a much more realistic picture of what humans had to navigate. That robotics by its nature involves dealing with uncertainty, with unpredictability, and so building a virtual intelligence wouldn't really illuminate the real problems of intelligence.

Re:Reminds me of MIT's AI lab

[The Milky Bar Kid]

Doesn't a Turing Test (which by its nature involves bits, rather than physical world) more accurately reflect the nature of intelligence?

Well, it depends on what you think intelligence is. The traditional view of AI was that the real meat of 'intelligence', the real tough stuff that we should try to solve, is rational reasoning. Playing chess, dealing with logic. Dealing with uncertainty and physically acting were not meant to be part of the problem - the view was that given a good enough reasoning machine, you could just plug on some decent sensors and motors and it would walk around just fine.

However, the reality turned out to be quite the opposite. It turned out that getting a computer to play chess is a hell of a lot easier than getting it to climb stairs. And so a couple of people, especially Rodney Brooks at MIT, started suggesting in the 1980's that everyone had had it the wrong way around - that dealing with uncertainty and unpredictability and the physical world was the essence of intelligence.

This school of thought is referred to as situatedness or embodiment. It suggests that intelligence is never as general as previously suggested - rather than a single great reasoning machine connected to motors and sensors, all the intelligent systems we know of - animals, us - are made up of a whole lot of highly specialized systems adapted to particular evolutionary niches. You cannot develop an intelligent system in isolation from the environment it is to work in - models always fall short. In Brooks' own words, 'The world is its own best model'. One of Brooks' points is that efficient land locomotion took much longer to evolve than reasoning - monkeys to humans is a shorter evolutionary step than fish to frogs.

So the end answer to your question is that a lot of people think, and there is reasonable evidence, that uncertainty and unpredictability are the real problems of intelligence, and understanding how a system develops and adapts to its environment is at the heart of discovering why and how we are intelligent.

And besides - robots are cool :)

Is anyone else afraid of these robots?

[mrroot]

I think we should enact restrictive legislation against the development of robots before it gets out of hand. If science fiction has taught us anything it is that robots will either...

1. Be incredibly useless
2. Provide comic relief

or, the one I'm concerned about...

3. Turn on humans, hunting us down one by one with unrelenting persistence.

Re:Is anyone else afraid of these robots?

[inerte]

Yeah, but it's like gun control. If you believe people can have guns, then you must assume there's a risk involved that things will go to one side, good, or bad.

I say we take the risk, and don't forget to put a red buton that reads "Eletrical Emission All Planet".

Re:Is anyone else afraid of these robots?

[Anonymous Coward]

Yes, we will all be pursued by shoe-box-sized battery operated robots, chasing us because we all emit the same frequency of infra-red light, desperate to drain our DC power.

Please.

Re:Is anyone else afraid of these robots?

[Whitehawke]

Please, go read ANYTHING that Isaac Asimov ever wrote on the subject of robots.

Serious science fiction (as opposed to Hollywood pap) pretty much gave up on the Frankenstein Complex fifty years ago; robots are *not* necessarily evil, nor are they necessarily interested in killing humans. Let it go.

Dave Storrs

University of Manga?

[GuyZero]

Wow! A whole university devotyed to the study of Manga???

Are they giant robots?

Do they fly on super-rockets?

Where can I enroll?

Soon to be overheard in the lab:...

[Mike Connell]

On a gallery overlooking the feeding pit ^H^H^H^H^H experiment lab...

TechA: "Aren't there meant to be 15 predators down there? I can only see 14"
TechB counts...
TechB: "Yeah, shit!", produces mobile, "I'll give Sharkey a ring..."
TechB, looking at mobile: "Batterys are dead. That's funny, I only charged them this morning..."

Insert dramatic exchange of glances and pause, followed by

AAAAAAAAAGHHHHHHHHH!!!!! Chomp! Chomp!

TechA in feeble voice "Agh! Number fifteen really is a bagbiter
TechB: It's, erm, sucking away my power dude!

etc etc...

This is not evolution

[pkaral]

This is not evolution, it is individual learning. At least at the robot-individual level, there are no random mutations and there is no death - both vital aspects of evolution.

Re:This is not evolution

[Atrahasis]

Yes there are random mutations. Yes there is death. RTFArticles

15 predators and 6 prey.

[Anonymous Coward]

This chap was on "The Big Breakfast" this morning (uk) and I'm sure he said that there was actually 20 or so "prey" and about 10 "predators"....

Quite interesting (ignoring the presenters inane comments)....but at one point the predator was "zapped" by the infrared from the camaraman's equipment.....

This isn't science.

[yehti]

After actually visiting the Magna site, I've found that this "scientific" experiment is actually a modified "Robot Wars" meets Vegas.

They talk more about the "dramatic music and lighting" than the design of the supposed experiment. Furthermore, who builds a custom arena to run a true AI experiment? This is just a publicity/TV ratings ploy.

Sounds more like a circus

[moniker_21]

"...spectacular 30 minute live action show - complete with atmospheric lights, smoke and music."

"Each show will begin in darkness. Dramatic music will flood into the arena as guests prepare themselves for the spectacular light, sound and science show."

Maybe I'm just a little jaded right now, but this sounds more like a circus show instead of a serious scientific experiment. I'm sure these are very complex robots, and the underlying idea is very interesting, but the whole BattleBots spin on it seems to trivialize the work. Now of course if he signs up Carmen Electra.......

I have heard of this.

[Em Emalb]

It's called battle bots. cheezy reference to stupid show=-1 over-rated.

Ok, so they are learning autonomous systems eh?

Great, how bout we let them learn something other than death and destruction?

Johnny-5 must hate hearing this news.=cheezy reference to stupid 80's show=+5 priceless

Re:I have heard of this.

[Atrahasis]

We're not teaching them it. They're learning it. And thats what animals in the wild do, and thats what our ancestors did. Its becoming increasingly evident in the field of AI that to achieve AI you have to start with something that isn't AI and let it learn.

Come on.

[Gannoc]

Will they fight back, will they run and hide? No-one knows, with each day, the robots change and evolve, and their actions will alter.

Yeah, like none of them has written a simulator showing what the robots could/will do until the year 4000AD.

If the prey could learn on its own how to "fight back", it would be an amazing acheivement in A.I, and you wouldn't need little robots runnnig around to demonstrate it.

Re:Come on.

[Atrahasis]

As far as I can tell, as I haven't seen the code, while my interest in AI and robotics really wants me to, there is nothing to stop them fighting back - thats a behoviour they can learn as well - if the prey learn to herd, then presumably, while the predator is chasing one prey, others could ram him in the side, and immobilise him.
The behaviour possible is almost limitless, assuming they haven't made any glaring errors in the code.

Re:Come on.

[Tipsy McStagger]

The prey would need a reason to do that though.. something like self awareness/desire to keep being or if the predators had lights underneath them that could feed some prey if they could flip it..

that'd be cool.

Re:Come on.

[Atrahasis]

Not really - the tendency could arise from the random changes in behaviour generated at each breeding. The problem would be that it depends on pack behaviour arising first.

If dolphins were monkeys

[JohnBE]

How's about we start gearing them up to kill or co-exist with something biological, maybe robots using horseshit as fuel, that'd be a natural environment.

I find the light interesting because it's something available in our natural environment, it'd be cool if the robots could use more of our natural environment to survive.

Survivalist robots! Let them loose in the woods and six years later they are talking about toppling the fedral government. Give them a Leatherman and in six generations they'd build a city.

You give CS a bad name

[rajeevishere]

50 years after Turing, and millions of dollars into AI research, we are still going in for Millenium shows of Intelligent colonies of Androids.
Sigh!

Audience participation is encouraged, the audience is asked to each pick a favourite, a pet to cheer for throughout the show, while the narrators are on hand to answer any questions.

This is outrageous. Cognitive Science as opposed to Good Old Fashioned AI was I think one of the most sensible currents in recent day CS research. And now just to discredit whatever sensible, realistic research occurs in these fields within the academic community here is Yet Another Crazy AI project grabbing front page visibility at /.

Prof..you give CS a bad name !

Re:You give CS a bad name

[Anonymous Coward]

"This is outrageous. Cognitive Science as opposed to Good Old Fashioned AI was I think one of the most sensible currents in recent day CS research. "

what is good old fashined AI? Are you talking about the symbol research? Are you talkning about the stupid Turing test, criticized in every introductory AI book?

This experiment is from the new direction in AI: Artificial life/genetical programs/Genetical algorithms/Neural networks.

"And now just to discredit whatever sensible, realistic research occurs in these fields within the academic community here is Yet Another Crazy AI project grabbing front page visibility at /. "

Thats the problem with I today: We all got an opinion about it, but nobody seems to understand what the AI community is working on.

Maybe YOU should do some AI research before being entitled any opinion.

Maintaining the entropy

[MadFarmAnimalz]

Hm, I'm wondering how the people in charge intend to maintain the sum of energy within the arena/universe. What happens if a spectator gets a little too attached to the prey and brings a high-powered flashlight?

This kind of resource balancing/world resource design scenario can get tricky... They ought to have called in Sid Meier :)

Re:Maintaining the entropy

[vandemar]

What happens if a spectator gets a little too attached to the prey and brings a high-powered flashlight?

I believe that would constitute divine intervention. Given the proper conditioning by the "gods", the robots may even change their behavior to earn favor from the gods. Naturally, there will be other robots that will lean towards agnosticism and atheism.

How the robots reproduce

[iapetus]

They 'reproduce' by taking the programs evolved by the more successful robots and combining them - pretty standard GP stuff, really. Those new programs are then fed back into the environment and allowed to evolve some more.

Solarbotics

[antisocial77]

Sounds like this dude took these bots and just modded a few to feed off the others. I encourage anyone who has an interest in robotics, even just a passing fancy, to check out that link. They are incredible fun.

Autonomic? What happened to autonomous?

[Aragorn379]

Were'd this autonomic crap come from? Gee, for the last 4 years I'd thought I'd been working on autonomous robots, going to conference with autonomous robots, etc. and know I find out they were autonomic. Oh well, I guess this means no Ph.D. for me. Anyone know how to break into the fast food biz?

Can you Imagine a Beowulf cluster of these?

[satanami69]

I don't even think the Discovery channel could get away with airing that kind of orgy.

Modular Robotics

[kippa]

The article in this month's IEEE Spectrum magazine, experimentation with modular robotics, seems more worthy of the label "ground-breaking."

Maybe not, but still valuable

[drew_kime]

Think of it like the team who found the Titanic. Roughly zero scientific learning, but the public interest in it brought in enough money to fund development of the remote vehicles. Once the cameras point to something else, they're left with some expensive new toys to use to do some real work.

Re:Maybe not, but still valuable

[gorilla]

There was archelogical interest in the Titanic. We discovered that it did break up before sinking in two distinct peices (The alternative theory being that it sunk almost whole), took samples which confirmed that the steel was brittle in the Atlantic enviroment, and even correcting the location of the titanic before it sunk, which answered questions about The Californian's ability to rescue Titanic passengers.

predator/prey ratio

[jd142]

Wow, they've really messed up the predator/prey ratio. Usually prey outnumber the predators by quite a bit.

Plus, the hype seems a bit Barnum-esque.

We'd be in big trouble...

[goldspider]

...if we combine this technology with that of the Meat-Eating Robot. :)

Why not simulations?

[crush]

Perhaps someone that works on this kind of thing can answer this question for me:
I can see that this might be a fun public spectacle and all but what, if any, are the advantages of building actual physical robots over doing simulations?
I am confused by the presentation of this as an experiment. I expect an experiment to be designed to potentially falsify an hypothesis. Is this designed to investigate predator/prey ecologies? If so, would it not be cheaper and more effective to write simulations or use some branch of discrete mathematics to model interactions with different starting values of parameters?
The only practical thing that I can see coming out of this is physical experience building robots, which is cool but is not what the aim of this "experiment" seems to be.

Re:Why not simulations?

[mikerich]

What's not been made clear is that MAGNA is a science museum with lots of hands-on exhibits. Physical robots are much more attractive to visitors than a simulation.

Perhaps one of the kids watching the robots zoom around will take some interest in AI and go on to do something more useful.

Best wishes,
Mike.

Re:Why not simulations?

[Anonymous Coward]

"... what, if any, are the advantages of building actual physical robots over doing simulations?"

"I expect an experiment to be designed to potentially falsify an hypothesis."

You answer your own question here. The experiment will prove that the simulations were (or were not) correct. Also the scientists do have a hypothesis: they think that some kind of cooperation will develop in the robots.

Simulation is fine, but it is only a substitute for actually building things. After all, a simulated engine won't get you to the store.

Re:Why not simulations?

[crush]

You answer your own question here. The experiment will prove that the simulations were (or were not) correct.

Well, _what_ hypothesis? It doesn't make any sense to talk about the simulation being "correct".

Also the scientists do have a hypothesis: they think that some kind of cooperation will develop in the robots.

Ah, now that sounds closer to a hypothesis, but it should be a little more clearly defined, otherwise we get into a situation where some physical behavior is observed and on the basis of that someone constructs a hypothesis and claims that it is "proved" by the data that went to construct it.

I don't mind robots collecting their own power...

[Archie Steel]

...as long as they don't build some that could make electricity from ingesting (and digesting) organic matter. That would make me feel uneasy...

"In the course of the last blackout, several people have been found dead after being attacked and half-devoured by their vaccuum cleaners..."

Quick Objection

[Dilly Bar]

I didn't read over this too carefully... BUT

I have have major objection to this. If they are trying to model natural systems, why do they have 6 prey and 15 predators??? In the real world a large prey population is needed to feed a smaller predator population. And while sometimes predator populations may get too big to support themselves, I highly doubt they will ever grow to over twice the prey population. Are they trying to model a system after a famine or disease wiped out the prey?

My prediction: The massive amount of predators quickly "kill" the prey and then if they can adapt quick enough, kill each other until one is left that eventually dies because it can't eat...

Re:Quick Objection

[alanwj]

I'd say that you are right on target about this.

I remember a picture from a science textbook way back in elementary school, labelled "Energy Pyramid". It had grasses and such at the bottom, with some rats on top of that, and on up until you got to things like eagles and such, which aren't normally consumed by other animals. It made the explicit point that as you move away from the primary producers into the secondary and tertiary consumers, a lot more energy per organism is required to sustain the population.

This is empirically supported if you just look at the relative bio-masses. What types of organisms do you see the most? Plants and algae. That is, primary producers, equivalent to "prey".

I am sure they are eminently more qualified to be conducting this experiment than I (if you can call it an experiment. It sounds more like a feature at Disney World to me.), but I would certainly like someone to explain how the expect 6 prey to stand up to 15 predators. It just doesn't make much sense. Perhaps they are just trying to make it something that will only last a day, for commercial reasons.

I'd also like to know (I didn't see anything in the article about it), if the predators are smart enough to go after each other when they get hungry enough.

Alan

Re:Quick Objection

[nerdlyone]

You assume that taking energy from prey kills it. It might only decrease the prey's energy temporarily, more like a parasite does to a host. A single host can take care of multiple parasites.

Why use hardware?

[Andrewkov]

It seems that these types of experiments would be much better suited to software than hardware. Building the robots, configuring them, etc, is time consuming and expensive. You can do simulations with software that can be exponentially larger (more robots) and much cheaper and faster to build and make changes later.

Object oriented design is perfect for this sort of thing. I did a simple experiment in Java, where predators, prey and food pellets were objects. Each object could have many different characteristics which chould be set when each object was spawned, which kind of mimics evolution. Also, if the logic in an object needs an upgrade (ie: The preditors are not too bright) it is easier to make modifications to the program instead of rebuilding a real robot.

I guess anything with real robots has a certain coolness to it, but any serious research in AI is better done in software simulations (not that I did any serious research, I was just learning Java and OO design).

Re:Why use hardware?

[Anonymous Coward]

"any serious research in AI is better done in software simulations"

I haven't actually read the article yet, so I can't comment on this particular experiment, but in general using real hardware gives you much more credibility. This is because it's much easier to make something work in simulation than on real hardware that's situated in the real world. In simulation you don't have noisy sensors, unreliable effectors, and boat loads of uncertainty about the environment. The environment also tends to be much less rich than the real world in terms of complexity.

I don't mean to knock simulation (I do research using both real robots and simulation) and some simulations do have things like a noise model, but the reaction I've seen a lot when people do something in simulation is: "That looks really cool. Now can you make it work in the real world?"

Simulation is a great tool, but since the gap between simulation and reality is so large people get much less excited about results from it.

cheers,
-Doug

Re:Why use hardware?

[jungd]

You are correct in that software is much more flexible in the ways you describe. Unfortunately, it doesn't work. Even the best simulations are nothing like the real world. Evolution has a nack for exploiting the difference. Which means that the software evolved on virtual robots stands no chance of running on real robots.

These experiments are old news. This has been done before in Brussels by Luc Steels et. al. Read some of that groups papers to understand why software evolution of robots would never work. (although I can still be a useful exploration of evolving intelligence in its own right).

Re:Why use hardware?

[jungd]

I couldn't find the link to their old robotic ecosystems projects, but here here is a link to their newer combo hardware/software system for language accusition through seeing. (click 'talking heads)

Re:Why use hardware?

[Black Parrot]

> It seems that these types of experiments would be much better suited to software than hardware. Building the robots, configuring them, etc, is time consuming and expensive. You can do simulations with software that can be exponentially larger (more robots) and much cheaper and faster to build and make changes later.

PR. Who ever heard of a press release about a simulated robot? We've got a planet full of AI researchers doing cool stuff with simulators, but you hardly ever hear about it.

The show-and-tell aspect probably helps land funding, too.

Re:Why use hardware?

[dollargonzo]

well, for one because it gets great press. software doesnt get much press. second of all, im sure they already TRIED using software and couldnt make enough hype about it. third, this is a ROBOT WARS judge! ofcourse he is going to use hardware. the article seems rather vague on what exact structure the neural nets have, and the info they DO give is kinda useless for experts and such since it is pretty much obvious how it is being done. hardware is more snazzy and makes more money, and looks cooler too. i mean c'mon, if yuo were doing this kind of experimint, wouldnt YUO want to do it w/ hardware too?

QED

Way to go, Slashdot Rocket Scientists.

[Anonymous Coward]

I've seen about 20 posts now saying "duuuh, not enough prey, too many predators."

Just like in real life, this is a FOOD chain. On a food chain, there are some animals that are prey AND predator.

Example: My cat eats mice. It is a predator in that regard. A fox would eat my cat. In that case, my cat would be prey.

The robots in this called "prey" are ones that can ONLY feed off the light trees. Some predators feed on them. Other predators feed off those predators, and so on.

Bunch of freakin' rocket scientists! ;P

Re:Way to go, Slashdot Rocket Scientists.

[Ratsarse]

This still violates the laws of thermodynamics. First cut: the total energy available to the system is that fixed by the prey. Some of this is used by the prey themselves. Some (most of it actually) is lost in the process of energy transfer. The further up the food chain your super predator is the worse its efficiency, about an order of magnitude efficiency drop per step up the food chain is ballpark. If they had paid even a lip service attempt at modelling a real system the problem would be even worse, with significant energy "lost" in growth, reproduction, and homeostasis. If the prey are unnaturally efficient at fixing energy, and if the predators don't kill the prey when feeding, then the system could run through several "energy cycles" before running down, but then the predators would be parasites. All this is not rocket science, its high school ecology, and only half a step sideways from accountancy.

There is absolutely nothing to do with evolution in this "experiment" (ie the measurement of change in gene frequency at a locus within a population over time measured in generations). There is no thesis, no null hypothesis, no intent or purpose at all. Its just a fishing expedition with both eyes firmly on the publicity. Its damn sad that crap like this draws funds away from real work.

Disappointment

[Anonymous Coward]

Fighting Robots? Manga? ... oh.. Magna.. =(

Would it be possible to change prey into predator

[Anonymous Coward]

Just wondering a little here,
What if as the 'experiment' progresses, starvation
conditions are slowly created.
It would be kinda cool if the prey-bots (if provided with the ability) started hunting their own to survive, rather than seeking out those light food sources. And perhaps more interesting would they develop techinques to save energy.

The ability to change one's food source would through something interesting into mix.

It would be neat to see these robots in action after they had several months of experience pitted against each other...

Not scientific at all

[pug23]

Calling this an "experiment" or "science" is complete BS. This has no value. Anything you could learn from a robotic simulation of that nature, you could learn far more efficiently from a pure software simulation. This is clearly entertainment masquerading as science. Now, that doesn't mean it wouldn't be interesting to watch, but, for crying out loud, don't call it science!

Server SlashDDOS'ed, here's the story...

[bob@dB.org]

Predator and Prey Robots set up home at Magna

World-first Living Robots show set to open at Magna in March 2002

ARTIFICIAL intelligence machines, cyborgs, androids or replicants, call them what you will. But free-thinking, independently acting machines have captured the imagination of authors, film-makers, artists, the military and governments for a long time.

From 27 March 2002, a colony of Living Robots, divided into 15 predators and 6 prey, will be will be on show at the Magna Science Adventure Centre, Rotherham's £46m Millennium Commission Lottery funded attraction.

'Living Robots' is a world-first experiment into artificial evolution - a culmination of eighteen months of research by world expert and Robot Wars judge, Professor Noel Sharkey and his dedicated team at the Creative Robotics Unit at Magna (CRUM)

The Living Robots have one goal, to obtain enough energy to survive and breed. The prey find their food from light sensors within the arena, while the predators feed off prey by stalking and chasing them before sucking away their power.

This groundbreaking experiment is being transformed into a spectacular public show at Magna. The amazing exhibition will take place in a purpose built arena, designed to hold 500 people at any one time.

In place of lectures and diagrams, the groundbreaking technology used in the robots will be demonstrated to Magna visitors in spectacular 30 minute live action show - complete with atmospheric lights, smoke and music. Guests will witness the robots in their natural environment, fighting for survival, learning and evolving as time goes on.

Each show will begin in darkness. Dramatic music will flood into the arena as guests prepare themselves for the spectacular light, sound and science show. Firstly, a 'prey' robot will be introduced - a good guy.

These smaller robots are powered by light and will automatically search the arena for special light spots to refuel. It is not remote controlled, but is full of computer chips controlled by an 'artificial neuron network' - a brain - and when it moves, it is because his brain tells it to.

Secondly another prey robot is introduced and the narrator will demonstrate how the prey can recognise friend or foe. This is done by an infra-red 'sniffing'. All prey send out the same infra-red light, different to the predators, and the audience will see that the prey robots have no instinct to run from each other but are happy to graze side-by-side under the light sources.

The one prey is then sent back to its pen. As the light dims a predator will enter stage. This is higher up the food chain than the prey, and survives by feeding from their power - the bad guy. The audience will be given a demonstration as to how the predators use its long tusks to entrap the prey and the sucking the power from the prey's battery. This is instinct - not remote control.

When the demonstrations are over, the show begins. All the predators and prey are released and from this point on, there is no control. Will they fight back, will they run and hide? No-one knows, with each day, the robots change and evolve, and their actions will alter. Audience participation is encouraged, the audience is asked to each pick a favourite, a pet to cheer for throughout the show, while the narrators are on hand to answer any questions.

The show will run throughout the day, times may vary. The show is included in the Magna ticket.

Yep

[Anonymous Coward]

According to Robert Bakker (sp?) - the bearded guy with the cowboy hat you see on every dinosaur show on the Discovery Channel - a balanced environment will have only 2% of the animals as predators.

Mod:Overrated?

[Anonymous Coward]

Dude, somebody's 'Overrated' button is sticking again. (like EVERY friggin comment is modded -1: overrated)

The Real Goal

[eric_aka_scooter]

Apparently, scientists are hoping to train a "breed" of robots capable of taking on the Battlebots. All of this got started when the scientists noted that the Battlebots TV show consistently beat out Science Digest in the ratings war. Never tick off a room full of PhD's...

Yeah....

[sonicattack]

The Living Robots have one goal, to obtain enough energy to survive and breed.

Sounds a lot like my own life.
Gotta get some more of that breeding, though.

Nothing more than a stage show

[Anonymous Coward]

with no real scientific significance. The methodology is way off.

Most obviously, you need far more prey than predator in the "ecological" system they propose.

Secondly, there needs to be far more of both types of robots in the system, and they need to have some minimal learning abilty. For instance... take over a whole building, placing feeder stations in random locations at a level where the prey can get to them. Make the prey different in design from the predators as well, so that they can find refuge in certain locations where predators cannot go. Make the prey somwhat competitive or territorial as well, so that they all don't end up packing into proteted area's.

Both predator and prey should breed (sharing their learned AI behavior and passing it down to their progeny), with new robots being introduced upon successful completion of the breeding cycle.

Make the feeder stations for the prey have finite resources... meaning that it can only put out so much food (power) each day.

Program two entirely different base sets of "instincts". Prey animals look for food and safe area's, congregate together in packs for protection, and seek refuge area's. Give them a docile pack mentality so they form bands and don't accept outsiders (other than breed "babies").

For the predators, make them solitary and aggressive (canibalistic)to other predators except when they need to mate (arbitrary cycle of time).

Make the prey have a minor defense from predators from the front only (as a programmed behavior). When attacked from the front by predators, the predator is stunned and looses some energy and has to learn to attack from the rear.

Give both a power management scheme as their central programming that hooks into their behaviors. They learn through this as part of tehir AI since their goal is twofold to breed and survive by acquiring energy.

I am sure there is more... it just seems that the experiment is really more of a press deal than anything else. Give us real science.

I am pretty sure that building these robots with complicated behaviors (as listed above) is technically not feasable just yet on a minor budget and due to programming an size constraints. However, the results would be interesting to see.

Re:Nothing more than a stage show

[Atrahasis]

Have you actually read the articles?
Most of your objections are groundless, and your proposed programming of complex behaviour, while possible, would defeat the object of the study.

They do have learning ability. They do breed. They are different in design.

Some other interesting work

[quantaman]

Another roboticist, Mark Tilden(http://www.wired.com/wired/archive/2.09/til den.ht), actually builds robots that have no CPU's. He fashions them after insects by having just a simple circuit board, after an action proves unsuccessful it gets changed slightly and like this the robots learns. I heard about one experiment where he took a number of solar powered robots (built out of things like walki-talki parts) that were programmed (I'm not sure of all the details) to find light and set them in a room with a few light sources. He observed behavior that some of the larger ones broke smaller robots and ended up using their parts to form a barrier around the light source.

Re:Some other interesting work

[quantaman]

Sorry, I was rushed last time and didn't get a chance to get the link working here's a link to some wired interviews (I coudn't link directly to one) and here's some pics of the actual robots.

Robot Wars

[doob]

For those who don't know, Noel Sharkey judges the UK Robot Wars. I wonder if he's borrowed any of the house robots for the predators :)

My professor.

[Placido]

No way! Noel Sharkey was my professor at Sheffield University! He also judges the UK Robot Wars. And that's all the useless information I have for you today.

Wahey, I feel famous. ;-) Blatant Karma Whoring (suckey, suckey?)
Noel Sharkey

Homepage

Google Cache

I think the Comp Sci web server is down so you should probably check the cache first cause you don't want to slashdot it.

EVOLUTION != SURVIVAL

[johnrpenner]

the RESULT depends on the goals you DEFINE:

'The Living Robots have one goal, to obtain enough energy
to survive and breed.'

thus, it is not like evolution at all, but comes with
a built-in BIAS that DEFINES their evolution.

"Think again before postulating the drive to self preservation
as the cardinal drive in an organic being. A living thing desires
above all to vent its strength - life as such is the
will to power - self preservation is only one of the indirect
and most frequent consequences of it". (Freidrich Nietzsche)

Re:EVOLUTION != SURVIVAL

[Alsee]

to survive and breed.

it is not like evolution at all, but comes with
a built-in BIAS that DEFINES their evolution.

Umm, can you give me any example of "real, unbiased" evolution that doesn't boil down to "to survive and breed"?

-

Re:EVOLUTION != SURVIVAL

[Sprunkys]

Well, survival leads to evolution, in the sense that evolution is the result of survival. Evolution is the change in a being that is preserved in future generations due to the being's (extended) survival.

it is logical that an individual wishes to survive and it is most common that individuals wish to breed.
Why is it logical to wish for self preservation?
Let's first look at an individual that does not live in a tight group (e.g. cheetah, humans) and ignore self-sacrifice (sacrificing one's life for someone else). It is then programmed in every individual to avoid danger and rather stay in a safe place and to look for food/energy rather than starving. This we call self-preservation, the word "self-preservation" is perhaps not what the creature has in mind when it flees from high water, dangerous temperatures or when it goes out to search for food, but it is effectively what it is doing, it is preserving it's own life
We can extend the problem by introducing group-behaviour, ants will sacrifice themselves for the benefit of the nest, humans tend to be heroic and throw themselves in front of bullets and such, but all this always relates to the preservation of genes, it is quite logical: a mutation occures that makes an ant sacrifice itself, this has a positive effect on the survival of the related individuals (thus the survival of similar genes) which means that the mutation lives on.

The same goes for breeding, if a mutation occures that increases the breeding for a certain individual it's genes will be more common and more breeding will occur thus in the end we will see that a specie breeds a lot. one than says that it is due to evolution, perhaps we should say that it is simply logical. the word "evolution" has the tendency to compete with "divine intervention" while it is pure logic that dictates the situation.

I find it very hard to think of matters as evolution and creationism, perhaps this comment will help you, o reader, to transform your thoughts and ideas into words, i would be very interested to hear them

Re:EVOLUTION != SURVIVAL

[dollargonzo]

technically, there should be a formally defined system of punishment and reward, and survival quite clearly fits into the category of reward. however, in the case of robots, it is very difficult to define the punishment and reward except by self preservation. also, no one REALLY knows (including nietzsche) what we strive for, and the best thing we have thought of (as good as any) is self preservation. yuo say power is the goal? what is power. that seems quite a bit more vague than survival. how do u define power in humans, even, for example...and more importantly in this case, robots?

QED

Re:already been done (more than once)

[Anonymous Coward]

Both articles are /.ed, so if I repeat, I apologise in advance.

But, the first I recall reading about predator/prey robots in this exact configuration was done by Mark Tilden, pioneer of Analog Robots.

Sure, it's been done, but so what? All scientific breakthroughs require repeated reproduction of results, otherwise, it's just some crackpot in a basement.

Stupid experiment set-up

[technophiliac]

Read the article and it sound more like a publicity stunt than a real experiment.

They ignore the fundamental rule in predator/prey relationships: there's more prey than predator ...

It's that triangle theory we all learned in high school biology.

When prey is outnumbered by predator, prey disappears, and then eventually predator disappears.

You must have blinked for a long time

[UberQwerty]

in order to miss where the selection ocurred. Only the surviving bots reproduce. They were selected in that they survived. Survival rate as a measure for fitness is not too crude, and this can be deduced by the fact that survival rate is the only measure of fitness. Well, that and ability to reproduce. But what you're doubtless thinking is that there are other fitness carachtaristics - like size, copmlexity, intelligence etc. These are survival charachtaristics only when they are means for survival, which they sometimes aren't. The only thing that an animal always wants is to survive. Evolution is not unilineal.

now every surviving bot has the same amount of fitness (offspring). That seems to be some binary kind of selection which I at least have never come across in real life

On the contrary, that's exactly what happens in real life. Indiviuals which survive either Reproduce, or Do Not Reproduce. And in a lot of animals, like, oh, say, humans - the number of offspring per reproduction is quite binary - 1 or 0. The fact that a human can produce more than one offspring in a lifetime is based on its multiple reproductions. Which, by the way, the robots also have.

Just like work...

[dbretton]

'The Living Robots have one goal, to obtain enough energy to survive and breed. The prey find their food from light sensors within the arena, while the predators feed off prey by stalking and chasing them before sucking away their power.'


ugg...must feed from flourescent lighting....
must not let pointy-haired boss get me....

interesting, but...

[Anonymous Coward]

this doesn't seem like evolution to me.

So their programming changes, but their bodies don't? It's probably much, much easier, for a given environment, to have a physical change than an increase in intelligence to survive.

Temp drops, say, what's easier - grow a thicker coat, or evolve the brains to harness fire?

Still, if they evolve any interesting behaviors, I guess it might be cool in the end...

Re:interesting, but...

[Atrahasis]

Their intelligence doesn't increase. Their ability to perform tricks to survive does.
They don't "know" they're performing the trick to survive, they just do it because thats what programming/instinct tells them to do.
Crickets/grasshoppers have been performing acoustical analysis for millions of years that humans were unable to perform until very recently - does that make crickets more intelligent than humans?

I know I've seen this movie...

[chip_s_ahoy]

Predator & Prey. Great title.

Professor Sharkey The well-meaning, but ill-fated scientist.

Guests will witness the robots in their natural environment. The bored, idle rich audience. The bored, idle rich, unsuspecting audience.

locked in an arms race that must end in stability or extinction. The dramatic situation.

The Predators have to capture the prey, immobilise them, and then extract their battery power with an energy-sucking fang that is stuck deep into the middle of the Prey. Vampires. Mutant Robot Vampires.

The experiment goes awry, with the robots turning on the audience, then being unleashed on the world. Uber-hacker Ariana Richards (or maybe Sandra Bullock) has to interface with the robots ("Oh, I know this, this is UNIX!") and distract them while Keanu has uses the robots one weakness in a race against time to save the planet. Paul Reiser gets killed try to capture one of the Mutant Vampire Robots for the defense industry.

Re:Damn!

[Anonymous Coward]

We are evil, because I just sold my soul to santa!

Learning autonomic nerds

[iangoldby]

'Drivelling Slashdotters' is a world-first experiment into artificial evolution - a culmination of four and a half years research by CmdrTaco and his team. The colony is divided into two species: Moderators and Posters. The Posters have one goal: to obtain enough karma to sound off about anything without being moderated into obscurity. Moderators feed off Posters by stalking down trolls, first posts, and anything vaguely factual, and sucking off score points.

In place of erudite discourse, the groundbreaking technology used in slashdot.org will be demonstrated to guests in a spectacular live-action show - complete with more-heat-than-light, flames, and random musings. Guests will witness Slashdotters in their natural environment, fighting for survival, learning and evolving as time goes on. Or not.

Each show will begin in darkness, and remain in darkness. Initially, a 'First Post' will appear. This simple creature is controlled by an 'artificial moron network', and it posts for no apparent reason.

Then a Moderator will appear. This is higher up the food chain than the Poster, and survives by feeding its ego off the Posters. The Moderator will immediately mod down the Poster. This is instinct, not reason.

And so the show goes on... and on...

(No offence intended...)

good laymen book

[aCapitalist]

For those who are interested, check out the book "Artificial Life". It's not heavy on the tech, but is interesting, and is good on the history of artificial life type stuff. I forget who the author is.

It's interactive!

[Alsee]

Audience participation is encouraged, the audience is asked to each pick a favourite, a pet to cheer for throughout the show, while the narrators are on hand to answer any questions.

Attempting to lable everything &LT Buzzword&GT interactive &LT /Buzzword&GT when is isn't is getting pretty annoying.

Oh, wait, I'm sorry. They never actually said &LT Buzzword&GT interactive &LT /Buzzword&GT. It's &LT Buzzword&GT audience participation &LT /Buzzword &GT. My bad.

-

Re:It's interactive!

[Atrahasis]

You're still right - how are the audience participating by observing and screaming at robots which have no way of knowing they are there?
The only way to interact would be to stand t the sidelines with an IR to confuse their recognition systems, or with an ordinary light to feed the the prey with "mana from heaven". Maybe we can set up a belief system....

Ground breaking?

[a_d_white]

This is a "ground breaking" experiment? Hello? Stefano Nolfi and Dario Floreano have already done essentially the same thing with a couple of Khepera robots. The only real substantial difference that I see is that this experiment is on display for the public.

References:

Floreano D. and Mondada F. (1996) Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics---Part B: Cybernetics, 26(3):396--407

Nolfi S. and Floreano D. (1999) Co-evolving predator and prey robots: Do "arms races" arise in artificial evolution?, Artificial Life 4 (4):331--335

Drones in a bee population

[tve]

Well, not strictly an example of the type you ask for, but only if you remember that it is the genes that are pushing for survival, not the individuals that carry them.

Re:your sig

[Anonymous Coward]

The plural of "CD" is "CDs". The possessive is "CD's". Figured you might want to know that..

Research project

[Eythian]

I am looking into doing something involving genetic algorithms as a research project. One idea that was proposed that, the more I think about it, the more intriguing it becomes, is the concept of geography. That is, having distinct sub-populations that rarely cross-breed. I'm thinking about implementing a simulation of this, perhaps with an environment that is similar to one that these real robots work in. It would be interesting to see if several subpopulations that rarely leak across would introduce severly new and effective tactics.

Re:Research project

[Atrahasis]

Geographical isolation is one of the most powerful driving forces in speciation and evolution, so even if you make the environments exactly the same, you can't expect the same behaviour to arise. It might turn out that it does always turn out the same, but the point is that evolution is an emergent phenomenon - you can't say whats going to happen without running an exact simulation, in which case you haven't made a prediction.
I am also very interested in computer simulations of this sort - I've already exhausted (my tolerance) of variations of Conway's life, and am now looking into the same sort of thing you are talking about. Good luck.

idea: distributed screensaver alife antfarm

[memeomatic]

Creatures, a book on the making of an alife game, was reviewed on /. a while back. I gave it a read, and started daydreaming about some cool applications.
Ant farms are fun - a screen saver that shows alife creatures of one sort or another going about their lives would be fun too.
Watching colony members battle it out and evolve over generations would be tres cool.
Making the application distributed in some way, so that there is migration of your alife to someone elses pc/colony would be cooler still.
Publishing an api so that users could code their own alife would be the coolest yet. Imagine coding your own creature, spawning a few copies of it, and releasing it into the great, wide net. It could battle for supremacy on the screens of other users, and someday, one of its ancestors could wander its way back onto your screen, perhaps evolved in a way you'd scarcely recognize.

Quick, someone go code it - i'd buy it.

Re:idea: distributed screensaver alife antfarm

[Fastolfe]

The network-aware Tierra was supposed to do a lot of this type of thing. You'd network a bunch of Tierra systems together, and the organisms could call a certain function to cause them to migrate to other systems on the network.

Unfortunately, like most other University-spawned projects, this project looks like it died as soon as the thesis was written.

I've tried implementing some stuff like this on my own, though, but I never seem to have enough spare time to finish it. A flexible engine with a published API and code, and sufficient opcodes for the organisms to actually do interesting stuff is what we need...