Slashdot Mirror
Evolving Robots Learn To Prey On Each Other
quaith writes "Dario Floreano and Laurent Keller report in PLoS ONE how their robots were able to rapidly evolve complex behaviors such as collision-free movement, homing, predator versus prey strategies, cooperation, and even altruism. A hundred generations of selection controlled by a simple neural network were sufficient to allow robots to evolve these behaviors. Their robots initially exhibited completely uncoordinated behavior, but as they evolved, the robots were able to orientate, escape predators, and even cooperate. The authors point out that this confirms a proposal by Alan Turing who suggested in the 1950s that building machines capable of adaptation and learning would be too difficult for a human designer and could instead be done using an evolutionary process. The robots aren't yet ready to compete in Robot Wars, but they're still pretty impressive."
What could possibly go wrong?!
Minor detail perhaps, but as Academic Editor in Chief of PLoS Biology I want to point out that the paper was in PLoS Biology not PLoS One ...
In time you can expect these robots to develop useful behaviours. Not those of most interest to grad students. Such behaviours will include microwaving food and contributing to slashdot.
And those who aren't saved go to robot hell, and must play the fiddle and beat the robot devil in order to leave.
Hey now. I for one welcome our new robot...oh fuck it. It's been done too many times.
There should have been a 4th law :-
Don't harm another robot unlessspecifically ordered to do so by a human.
Clearly, these mechanical creatures were designed by a higher intelligence.
This in no way confirms that it would be too difficult for humans to build robots that posses higher A.I. traits, nor does it confirm that evolution is a better process than intelligent design.
actually, as of now, these robots are just programs in a physics computer experiment... so if they were to evolve to be smart, we'd have a computer virus instead of an actual robot that is evolving. I wonder, if a robot program like this were let loose on the internet, and was capable of learning... what would it learn?
do they believe in God?
Skynet will be evolved in JavaScript.
Deleted
" the robots were able to orientate" ... Neat. Wow. Did they have an internal compass? Orientate means to "face east", specifically, toward Mecca.
worth RTF'ing for a better idea of how this is done (btw, they are the same robots that were taught to "deceive" other robots about where the "food" is). Plus, the video of the Predator/Prey stalemate is just epic! As for the 3rd video (maze navigation), man, i would have blown these 1st gen robots to pieces before they could say Darwin!
I've always thought that "Real AI" wasn't something we could design, but would need to evolve to the point of intelligence. We already know it works, it's just a matter of application.
What if this was allowed to span not 50, but 50,000 or 50,000,000 generations?
Now imagine all the time it took us to evolve in that capacity and do it in the span of a few minutes.
I think the ability to have AI is already solved by today's hardware; we just need the right kind of software.
Definitely an interesting continuation of work being done by various groups over the past couple of decades.
But one thing to note is that crossover isn't especially useful in neural network evolution. In early stages of evolution, it's really no better than random large perturbation of large swaths of the genome. In later stages, it can actually decrease the speed of evolution toward high fitness genomes, because at least some of the time (particularly if there are multiple "species" in the population) crossover ends up being a random large perturbation which hinders the search of local fitness space by mutation; the rest of the time (when individuals from the same "species" are crossed) crossover is no better than mutation.
The reason for this is because the parameters of a neural network are not functional. A section of the genome may correspond to a weight between neurons, but that weight doesn't have a specific function. In biological organisms, each gene is transcribed/translated into a protein, and that protein may have a particular function within the cell. If that gene is acquired by a descendant through crossover, the protein could serve the same (or a somewhat modified) role it served in its parent, even if the rest of the descendant's genome was acquired from the other parent. But with artificial neural networks, the parameters were all evolved as parts of a whole, where each individual parameter has no function on its own, but the behavior emerges from having all of those parameters at the same time.
This could potentially be mitigated by the genome encoding scheme one uses, and of course, if the crossover rate is low enough, the ultimate effect would be small.
DIY manual by Daniel H. Wilson on how to survive the coming uprising.
The noun "orientation" is derived from the verb "orient", not the other way around.
-- "At Microsoft, quality is job 1.1" -- PC Magazine, Nov. 1994
The predator and prey bots reminds me of sales people chasing around after anyone who wanders too closely while they try their sales pitch.
This kind of behavior was first demonstrated/modeled (AFAIK/IIRC) as part of the Tierra simulations almost twenty years ago. Though I don't have a reference to hand, I know it's been done in neural networks before too.
So other than the 'sizzle' (as opposed to 'steak') of doing it with robots, can anyone explain what is new here?
According to TFA, the robots were controlled by neural networks, not the selection process.
Spoiler for the story - since it's basically the ending - but the point in question:
As the Tasso models approach, Hendricks notices the bombs clipped to their belts, and recalls that first Tasso used one to destroy other claws. At his end, Hendricks is vaguely comforted by the thought that the claws are designing, developing, and producing weapons meant for killing other claws.
It's worth noting that the robot in the experiment where they evolved the bodies as well, in order to run faster, looks like a shark, with a tail and 2 fins. Fascinating. If they could do the same experiment with the ability to walk instead of only crawling (see video http://www.plosbiology.org/article/fetchSingleRepresentation.action?uri=info:doi/10.1371/journal.pbio.1000292.s006 to see what i'm talking about ) and make them do something that requires hands such as lifting something up, we could see if the optimal forms were humanoid, centaur-like, spider-like etc.
... everything about the experiments is setup and designed, no real world intelligence evolved like this. This is more like being in control of the weather, causing it to snow, laying the snow just right, then rolling up a big enough ball of snow and letting it down the side of a mountain.
You don't need physical robots running around a maze to demonstrate AI.
Seastead this.
A new meaning to next....
Surely the robots were themselves controlled by neural nets which were selected by Genetic Algorithm, rather than using a neural net to control the selection process itself. Perhaps if I RTFA...
Can you be Even More Awesome?!
orientate
v : determine one's position with reference to another point
[syn: orient] [ant: disorient]
Over-the-top Response Guy! Giving "Over-the-Top Responses" since 1970.
This doesn't "confirm" anything about Turing's offhanded opinion.
Strange things are afoot at the Circle-K.
So, a group made some beings in a 'universe' that are unable to see their creators. The beings were made self replicating, and were fiddled with. Then group withdrew and left the beings to their devices. Eventually beings denied being created in first place.
Where have I heard this before? This is eerily familiar...
skynet, wopr, or maybe the cylons
If you want to do this right then have your robots spend more time with humans than other robots. This way we can evolve a robot who plays well with other people, than with other robots.
That's what I'm sure my favorite robot SF authors -- Elf Sternberg and D.B. Story -- have planned for their robots. I would love to meet either of their creations.
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
Your point is valid if the genotype to phenotype mapping is a simple mapping to neuron type, connection type, weights, etc. However, we clearly have effective crossover in humans, which means there can be a genotype to phenotype mapping that operates at more of a functional level. It's an interesting and difficult problem.
I thought you might have male and female algorithms for some optimization problems "walking around" in a virtual world, mate to create combined algorithms, giving you some kind of blood-relationship between the algorithms (and yes, they should die after some time). the related algorithms would form "clans" and if males of opposing clans meet, they fight over each others ressources (RAM and CPU-Cycles) and there should be sources of these ressources, which should dry out over time, so the algorithms HAVE to migrate and attack each other to get new ressources... and fighting should drain some ressources (but having more ressources should give you an advantage in a fight, like being allowed to run longer or use more RAM)
having more ressources should make the male algorithms more attractive to the female algorithms and there should be some different kinds of "agressiveness"...
this is exactly how nature evolved our brains - this should really work well for genetic algorithms...
ultimately when one algorithm owns all the ressources, he's "the result"...
I only fear that the algorithms might develop some p*ssy features like compassion, culture, science, sharing, etc. This would make the results weak and useless! I'll have to hard-code religious fundamentalists, rabble-rousers, RIAA lawyers and republicans! And I have to give different clans different commandments so they can fight about who's commandments are the commandments by the one, true GOD!!! MUAHAHAHA
and god damn, I'll have to give a name to the ressources... I might call them spice... or oil...
The MAFIAA is a bunch of mindless jerks who will be the first up against the wall when the revolution comes
get a emp gun!
Evolving robots learn to prey on each other.
They're calling them [insert despised political party here].
Robots might orient themselves but orientating themselves must involve eating potatos while finding their directions.
Are they connected to the internet?
Just asking, cause I might need to start escaping now, you know?
A simulation I developed around 1987 had 2D robots that duplicated themselves from a sea of parts. They would build themselves up and then cut themselves apart to make two copies. To my knowledge, it was the first 2D simulation of self-replicating robots from a sea of parts. The first time it worked, one robot started canibalizing the other to build itself up again. I had to add a sense of "smell" to stop robots from taking parts from their offspring. As another poster referenced, Philip K. Dick's point on identity in 1953 was very prescient:
http://en.wikipedia.org/wiki/Second_Variety
"Dick said of the story: "My grand theme -- who is human and who only appears (masquerading) as human? -- emerges most fully. Unless we can individually and collectively be certain of the answer to this question, we face what is, in my view, the most serious problem possible. Without answering it adequately, we cannot even be certain of our own selves. I cannot even know myself, let alone you. So I keep working on this theme; to me nothing is as important a question. And the answer comes very hard.""
However, those robots were not evolving. I presented a talk on that simulation at a workshop on AI and Simulation in 1988 in Minnesota, saying how hard easy it was to make robots that were destructive, but how much harder it would be to make them cooperative. A major from DARPA literally patted me on the back and told me to "keep up the good work". To his credit, I'm not sure which aspect (destructive or cooperative) he was talking about working on. :-) But I left that field around that time for several reasons (including concerns about military funding and use of this stuff, but also that it seemed like we knew enough to destroy ourselves with this stuff but not enough to make it something wonderful). At the same workshop someone presented something on a simulation of organisms with neural networks that learned different behaviors. A professor I took a course from at SUNY Stony Brook has done some interesting stuff on evolution and communications with simple organisms: :-)
http://www.stonybrook.edu/philosophy//faculty/pgrim/pgrim_publications.html
Anyway, in the quarter century almost since then, what I have learned is that the greatest challenge of the 21st century is the tools of abundance like self-replicating robots (or nanotech, biotech, nuclear energy, networking, bureaucracy, and others things) in the hands of those still preoccupied with fighting over percieved scarcity, or worse, creating artificial scarcity. What could be more ironic than using nuclear missiles to fight over Earthly oil fields, when the same sorts of techology and organizations could let us build space habitats and big renewable energy complexes (or nuclear power too). What is more ironic than building killer robots to enforce social norms related to forcing people to sell their labor doing repetitive work in order to gain the right to consume, rather than just build robots to do the work? Anyway, it won't be the robots that kill us off. It will be the unexamined irony.
A 21st century issue: the irony of technologies of abundance in the hands of those still thinking in terms of scarcity.
---
Robotics Feed @ Feed Distiller
I can never understand why experiments like this use robots, when all the robots actions could be easily encapsulated in software. I think this vein of research may lead to some interesting mechanical advances, but real AI progress will happen much faster by people using more efficient techniques.