Slashdot Mirror
A Robot Learns To Fly
jerkychew writes: "For those of you that read my last post about the robot escaping its captors, there's more news regarding robots and AI. According to this Reuters article, scientists in Sweden created a robot that essentially 'learned to fly' in just three hours. The robot had no preprogrammed instructions on how to achieve lift, it had to deduce everything through trial and error. Very interesting stuff."
damn you :P
Not only do we have to watch out for bird crap raining down on us, we now have robot excrement to worry about as well.
~~~~~~~~~ "I must create my own system, or be enslav'd by another man's." William Blake, Jerusalem.
It just seems to me like AI through logical progression, which I'd be tempted to not call AI...
The fact that it "cheats" somehow restores my faith in robotkind....
-ajb
A robot has taught itself the principles of flying -- learning in just three hours what evolution took millions of years to achieve
Well. Assuming the birds were TRYING to fly, knew what lift was, and already had the equipment (i.e. wings) to achieve this.
This brings an image of stupid birds sitting around flapping randomly thinking "FUCK - I'm SURE this should fucking WORK! - Bastards - OOps, I just fell over to the left - does that mean my right wing was flapped right???? - Hey - John! WHAT DID I DO THEN????"
"However, the robot could not actually fly because it was too heavy for its electrical motor."
One small step for robot, one giant leap for robotkind
You'd think Reuters would be able to handle a little Slashdot effect, but apparently not...
what evolution took millions of years to achieve
CameOn camaleon, the evolution creates everything from the unicellulars in millions of year. This sounds like the press announces from some software companies.
Especially tried to cheat by standing on it's wingtips or similar. I would like to see something else though. What if we build lots of small generic robots, let's say they have wheels to move around only. The on the floor there could be more components that robots can attach themselves to, like giving them legs, wings, arms, eyes, ears etc., and then give them all different objectives, for example to survive, escape, learn from others, etc. Could be interesting to see if it would evolve into some kind of robot society where they all evolve different abilities and so on.
Will work for bandwidth
The moment the robot asks for a hamburger, hookes up to the net and orders a ticket for the next flight wherever they are getting somewhere with AI and simulated evolution..
Nobody expects the spanish inquisition!
Doesn't something sound just *slightly* wrong with that? Working it out as you go along... Hmmm, not sure if this is going to take off, (pun intended).
Imagine a day where engineers build cool robots, upload the generic learn-to-do-stuff-with-your limbs program, leave it for a week or so to train up and get optimum calibration, then have it copy it's program onto subsequent batches.
I picture a robot aerobics class.. heh. But if anybody asks, I picture a robot boot camp.
My life is one big siesta in which I'm dreaming I wished my life was one big siesta.
LONDON (Reuters) - A robot has taught itself the principles of flying -- learning in just three hours what evolution took millions of years to achieve, according to research by Swedish scientists published on Wednesday.
Ridiculous to compare prebuilt robot to evolution from some dinosaur to flying dinosaur (also known as bird). This really is tabloid headlining at it's purest.
And the robot didn't even fly, just generated some lift!
It's like saying humans can fly, when they generate 1N lift flapping their arms.
But it's great to see how selflearning robots and programs will start evolving now. I quess pretty soon computers and robots will be able to evolve faster on their own than when developed by humans.
It's fed a set of instructions, apparently 20/sec, and is asked to remember which one got it the highest.
Execute instruction
Lift higher than others?
YES - Remember this instruction
NO - Get next instruction
Repeat until no instructions
keep repeating successful instruction
Seems pretty basic to me and hardly learning, just a new spin on analysing the effeciency of algorithms.
Glenn
The Smrt way to trade CFDs on the ASX
You're all doomed, I warned you!
I'll just get to packing my stuff, moving to a remote cabin in Montana and keeping a close eye on my refridgerator (I know it hates me, it keeps melting my ice cream).
No boom today. Boom tomorrow. There's always a boom tomorrow. - Cmdr. Susan Ivanova
Its not really mimicing evolution now is it, birds developed there wings over millions of years, the robot had them to start with.. what would be kewl is to create a robot that has the ability to construct itself and then see what happens.
moo
what evolution took millions of years to achieve
Well, at least evolution succeeded in making birds that weren't too heavy for their own wings...
Seems to me that this project was not really as exciting as they would like us to believe...
This thing was programmed to learn how to fly...so what...it didn't just decide it would leave the ground on it's own.
You make it sound like it came to it's own conclusion.
The programmers did everything but give it the end parameters...it only needed to finish the math.
Big deal....
A 747 can land itself, and it's a heckofalot more complicated. I don't see any headlines on that today.
Rather than comparing this to millions of years of evolution, perhaps it would be better to compare it to a bird just old enough to physically be able to fly.
The robot was physically equipped with all it needed to 'fly'; it was also equipped with all the wires in the right places. The fundamental difference between robots and living organisms is in the thinking: a newborn bird has to forge new synapses in its brain; this robot was designed with the purpose of 'learning to fly', so was given all the appropriate connections; it is just a matter of working out what sequence of events is required. Robots inherently have some form of co-ordination; birds, on the other hand, just like any other animal, have to develop such skills.
Like car accidents, most hardware problems are due to driver error.
as everyone knows, it all depends on what software there was in the system. If the starting-point was a program, which contained instructions for trying to move the "wings", and seeing which instruction caused most lift, and tuning the algorithm based on that, I don't think theres anything fancy in it. If this is the case, this could have been done at the same time when the moonlander game was first done :) I mean, it all depends on how dedicated for this exact "learning purpose" the SW in that robot was - or was it just an self-optimizing algorithm. Is there any more details on the software inside somewhere?
Post 911 babble science to make us all feel warm and fuzzy. I mean, what the fsck?
Sorry but that article is just beyond me. And the robot tried to cheat!! Well yeah if it can turn millions of years of evolution into a three hour process it shouldn't have any problem learning how to cheat.
"Hey guys, look! We stood on really tall stilts, does this mean we're flying?"
That would have been something to see.
The robot stands proudly on it's wings, and tells the scientists "Look at me, I generated maximum lift, and I don't have to exert any force at all. Oh, and from here, I can see the mouse is climbing over walls to get to the cheese without going through the maze. You humans are so stupid!"
Whoever stated that signature sizes should be limited to one hundred and twenty characters can just go ahead and kiss my
I am not sure if you can call what the robot was doing 'flying'. It was essentially just flapping its arms in the most effective way possible with whatever wing-like appendages given to it.
Now the cheating - that is the interesting part. When they have the algorithm down so that the bot hobbles out the door and purchases a ticket at the airport, then they will have a winner.
"First you get the Linux, then you get the power, THEN you get the women"
check out AUVSI's Aerial Robotics Competition
Cheating was one strategy tried and rejected during the process of artificial evolution -- at one point the robot simply stood on its wing tips and later it climbed up on some objects that had been accidentally left nearby.
...
...
But after three hours the robot discovered a flapping technique
However, the robot could not actually fly because it was too heavy for its electrical motor.
"There's only so much that evolution can do," Bentley said.
Finally we understand the dodo's place in evolution.
**AA: a bunch of mindless jerks who'll be the first against the wall when the revolution comes
Here's what I did to play around with breeding algorithms from small building blocks:
Define a very simple stack-based language. The stack only holds boolean values, and when empty pops and endless supply of "false" values and when full discards pushes. Choose some control flow opcodes:
NOP, SKIP (pop, if true, skip ahead a fixed amount), REPEAT (pop, if true, skip back a fxied amount), NOT, RESET (clear stack, back to beginning)
and some opcodes related to your environment (mine was a rectangular arena):
GO (try to move forward one step, push boolean success), TURN (90 degrees clockwise), LOOK (push boolean "do I see food ahead?"), EAT (try to eat, push boolean success)
Pick a stack size (this has interesting consequences, as some of my organisms learned to count by filling the stack with TRUE values and consuming them until they hit the endless supply of FALSE when empty) and a code size. Force all organisms to end in your RESET op. Generate them randomly and run them in your simulator (I did 20-50 at once letting each one run a few hundred instructions in a row). Evaluate fitness (in my case, how well fed they were) and breed them. You can combine the functions in lots of ways. Randomly choose opcodes (or groups of opcodes) from each, possibly with reordering or shifting. Introduce some mutations.
Once you get something interesting, try to figure out how it works. This can be the hardest part -- my description above produced many variations that were only 8-10 instructions long before an unavoidable RESET opcode, and they could search a grid with obstacles for food!
MAIN
{
target = 72;
do
{
guess = rand();
}
while guess target;
print "GOT IT!"
}
NEWS HEADLINE:
Artificial Intelligence researcher creates computer program that comes up with the number 72.
The objective of the learning algorithm was to achieve maximum lift while attached to two vertical poles . So the headline should be: 'Robot learns to achieve maximum lift by flapping wings while attached to two poles'. I think keeping balance, avoiding stall, etc. are much harder to achieve.
That reminded me a quote from "Chicken Run":
Rocky: You see, flying takes three things: Hard work, perseverance and... hard work.
Fowler: You said "hard work" twice!
Rocky: That's because it takes twice as much work as perseverance.
This is complete and utter tabloidian rubbish. You can't compare the process of evolution (success through random mutation) to a pre-built machine that is given explicit instructions to overcome a programmed obstacle.
Evolution? That's the rubbish. They built the optimized wings for the computer, they gave it the "muscular" control and the "neural" connections to command them, then they think it's news when it "flies"? The order, structure and design of the universe screams for the existence of a sentient creator that planned and DESIGNED its inhabitants and environment. Birds were created with optimized wings, muscular control and neural ability. They fly because they were designed to fly. Think rationally -- how "useful" would a primitive wing that can't produce lift, can't grasp objects and can't forage for food have been for the species to have considered it a "beneficial mutation" to continue via natural selection?
So now we have robots that learn to speak, fly, sing, walk and talk. Well, monotonous, mundane and trivial tasks.
Really I wanna see the feat when a robot learns to be a sysadmin in a freaky corporate network, full of Windows servers/stations and having lots of lamers around him...
Interesting to see if BRfH (Bastard Robot from Hell) will be one the optimal solution to the problem...
amen!
Seem to be a lot of folks who aren't very impressed by this. I'll admit that the headline is a little over the top, but the story is still interesting -- and fraught with interesting potential.
For example:
StarBot: How long before it learns to make a grande latte half-skim/half 30 weight?
Bouncebot: How long before it learns not to turn its back on the loud drunk in the corner?
Lobot: How long before it decides it really doesn't want to learn anything, just sit around and smile.
Congressbot: how long before it learns that working tirelessly for your constituency is its own reward, whereas lying for assorted interest groups is money in the bank? Note: This may be a special case of the Lobot.
Rome taught me patience and assiduous application to detail. Virtues which temper the boldness of great, general views.
I study at this university :)
http://www.ida.liu.se/ext/witas/
C'mon this article doesn't even have a photo of the robot. Any robot can be built to eventually reach a predicted outcome. Of course it will flap since thats what it was given wings to do. This is as exciting as giving a robot a lift fan, except in that case there is one step to success, the "turn on lift fan" command.
What would be more interesting is build a robot with many different unique mechanisms such as fans, deflectors, arms, and whatnot and see if it can produce lift in an unexpected or unique way.
I don't proof read
This could lead to important developments in the future. The robot had a simple goal, but imagine if it was smaller, and it's connected to a bio-scanner that measures your "overall health". It could be programed to keep this health rating at a certain level, searching and killing what is making you sick.
Buy a Nintendo DS Lite
"It was amazing," said Dr. Heinrich Hienrichson, "Before I knew it, the robot had stolen my credit card, set up an account on Orbitz and booked two airline tickets to Mexico. Now the robot has escaped and my toaster appears to have gone misisng as well..."
----------------------------------- My Other Sig Is Hilarious -----------------------------------
Would anyone with some AI background comment on this story? It seems to me that since the robot had instructions on how to move limbs differently and the feedback mechanism - it didn't actually learn how to generate maximum lift - it just selcted the best combination of already installed istructions. Isn't this some sort of a simple ANN?
Without details the article is fairly pointless. If it actually flew the article would have a point as it is. Otherwise, it DOES sounds like just a GA. I'm going to the homepage of the professor to see what's new and different about this.
As others were, I was annoyed at comparing it to millions of years of evolution. Evolution had to make the physical design as well, it's test for success is a lot slower and fuzzier, and it's goals quite different.
An example of evolving the PHYSICAL aspects AS WELL AS the neural ones can be found in Karl Sims blockies. The little movie is pretty cool to watch. This still has the last two differences from evolution as the above, but there is no article saying the blockies are beating out evolution in 3 hours. blockies
(Note: I tried the link from the earlier post and it didn't work, guess I'll have to ask google.)
I find more interesting stuff in the codes at this place:
LEET war
where you're supposed to code a bot that kills all other bots in a game in a tournament. The AI stuff is WAY much more complicated than a stupid fly-robot that compare wingclapping with height achivement.
... they put 'Wings' on it. What kind of a challenge is that?
You fool! You've given cheese to a lactose intolerant volcano god! Do you know what that means?
Now quickly, robot, go rescue Amidala before she's covered by molten metal!
Garg
Garg
Alumnus, Xavier's School for Gifted Youngsters
What the researchers did was to build a robot that had wings and motors for manipulating them. These could be controlled by a computer. But instead of writing an explicit program telling the robot how to fly, they got the robot to learn how to fly. They did this using some sort of Genetic Algorithm.
Basically, what a GA does is to generate a large population of possible solutions to the problem, then evaluate how good each one is (i.e. measure the lift each one creates in this example) and then to breed good solutions to create successive generations of possible solutions which are (hopefully) better than the previous generations.
Then, once some criterion is met (for example, once the average fitness of your population doesn't change much for several generations), you then select the best solution found so far as being your answer.
In mathematical terms, GAs are stochastic methods of optimising a function; they are typically used when solving the problem using an analytic method would be problematic (i.e. it would take too long etc.).
So it's not really surprising the robot learned to 'fly' -- the researchers just managed to find an optimal sequence of instructions to send to the wings.
The next step would be to get a robot to learn how to hover without the aid of the stabilising poles; then fly from one location to the other; then fly in a straight line in the presence of varying wind etc.
What the research does do is to lend credence to the argument that insects and birds could have evolved, rather than having been 'designed' by some sort of a God.
"The noble art of losing face will one day save the human race"---Hans Blix
Hmm... I don't think this counts for much, personally. They had already built the machine, and optimised the outputs to control the motors efficiently, etc. That's doing most of the work, imho.
If they can get a robot to move sets of feathers in a complex, flexible wing in order to hover, and then get it to learn other flight techniques (such as diving), without screwing up the first things learned. Then, that will be very cool. And it might even be something we haven't known we could do for ages now.
A soul in tension that's learning to fly
Condition grounded but determined to try
Can't keep my eyes from the circling skies
Tongue-tied and twisted just an earth-bound misfit, I
(Pink Floyd)
I doubt, therefore I may be.
So we have this thing with wings. And lets just say for instance we write a program with a bunch of "for" loops even. This is probably quite high-level but let's say it loops through all angles of rotation of the wing and all upward/downward movement of the wing. These numbers can be put into a formula most likely. Which should end up eventually being the desired number that tells the robot that it did its job. Why did they even build a robot? All they really had to do was try every cobination of movements which then provided feedback which was most likely calculable. Am I missing something? It's not even so neat that it cheated... so the loop didn't tell the other wing to do anything... whoopie.
i really dig the idea of genetic "learning" simulations. they start with nothing, and eventually can come up with all the same things animals do - including different gaits for walking and running, etc.
this is especially cool, in that they've not only done this in a simulation, but with a real nuts and bolts 'bot. how easy it seems to me now to ship out robots with very little programming, but a quick learning curve. the owner puts the 'bot in its home, punches in a few things it would like the bot to do, and lets it explore a little. after some training, it's perfectly suited to its new job and new environment...
oh yeah. i grabbed a screensaver a while back from this guy that simulates a simple creature learning to walk. pretty spiffy, and you don't have to worry about it ambling off to the parking lot...
- Entertaining Bits from the Ancient Kernel Tree
Now they're working on a slashdotter robot. It sits in a comfy office chair across the room from a pot of coffee and has a limited power suppy. They feed it random instructions at 30 second intervals with the goal of replenishing it's energy level and optimizing the 'looking busy' variable.
This story can at best be equated to a hatchling bird that has come of age and is first learning to fly. Some birds may take days, while others take a few short hours. In no way has the machine developed the ability to fly, but much like the young bird has discovered it has the mechanical ability and has now atempted to put it to use in order to survive.
I hate all sigs, even this one.
where "nature" is the researcher(s). When the robot finds that it can't really fly, it could come up with an "idea" and ask "god" to lighten up its wings or motor, or have a faster motor, etc. That way, the machine would truly evolve as it goes. The researchers could even take these "requests" from the robot, and give it things it didn't ask for. For example, if it asks for a lighter motor, give it a faster motor instead.
August 16th, 2002 8:24am Skynet Goes online.
--Should work--
The really cool things about applications of GP/EA here isn't that it learned so much as the solution it came to. Often times, especially when we're thinking about ways to do stuff in hardware, rather than some simple software algorithm, evolutionary computation results in solutions that we humans wouldn't have thought of.
/. had a story about using FPGAs and EA to program an array of FPGAs to distinguish the difference between "Yes" and "No" (or something to that effect, perhaps on/off). After many generations, the solution it came to not only work, but why it worked wasn't understood by the scientist. The best known human solution for something like that would've taken twice as many circuits. It included factors and variables (like magnetic resonance given off by excluded FPGA chips that weren't part of the circuit, but were still recieving power! when removed, the circuit didn't work) that humans wouldn't normally consider.
A system that evaluates the effectiveness of a solution and refines it from there, that does so using real-world fitness ends up including factors that an engineer wouldn't have thought of. Perhaps because these variables are unknown, or seeminly insignifigant.
A while back, you'll remember,
Man, I love this stuff.
Working toward a usable PDA environment in the spirit of Newton OS: Dynapad
"This tells us that this kind of evolution is capable of coming up with flying motion," said Peter Bentley, an evolutionary computer expert at University College, London.
Birdwatches around the world were SHOCKED at this finding.
However, the robot could not actually fly because it was too heavy for its electrical motor.
"There's only so much that evolution can do," Bentley said.
Using these definitions, this robot's achievement pales in comparison to my evolutionary process which lead my body to the ingest a jelly donut covered in sprinkles this morning. Since I never had one of those before, the only explanation is that I naturally evolved to the point where I wanted to attempt the ingestion of such an object.
Seriously, guys, this is nothing but cheap heat for a worthless techie. Move on.
This is what you get when news gets filtered down from a research report to a popular science magazine article and then, through a couple of clue-less journalists, to a sensationalist press release.
Here is the actual Research web-page
I miss my rubber keyboard.(Homepage)
These people just won't learn. What is presented as a breakthrough in the headlines turns out to be a VERY minor development. The little fellah already had all the basic elements - wings, more precisely. It just learnt how to flap them - not how to fly, for it is way too heavy for that. Measured by the same yardstick as natural evolution, such achievement is ridiculous.
Once again, a step forward in the AI field is blown out of proportion, therefore keeping this discipline firmly in its place of one that talks too much, and delivers too little.
Having been on a Robotics course, I knew what tripe this would be before I even read the article. The fact it was allegedly "learning to fly" is totally academic - it could have been trying to hit a football or move towards a light. All they are actually doing is producing negative feedback using semi-random programming. Damping the system to produce more and more desirable results as the process continues.
..... hence I confidently predict an IP claim within the week will be made for "Experienced based learning software and related hardware". You got it, they're gona sue all life-forms on the planet for infringement of IP! :)
:D ).
This isn't rocket science, it's not even particularly clever. It's the sort of thing they gave as routine lab work on the first year of my undergraduate Cybernetics course. And even then I thought it was pretty pointless. As many people here have already pointed out: Why even give it a physical body? It would be far more efficient to simulate the whole process in a computer environment. The answer is of course that robots get column inches.
If the robots reconfigured their physical bodies as part of this life-cycle (like the replicators can in Stargate) - then you would have the beginnings of a evolutionary process.
I would even dispute the idea that the robot was learning. Learning implies that it considered various facts then made a decision based on those facts. From the descriptions given (both above and in the article) the code consisted of starting with a random seed of commands - mutated it in various ways, then selected ones that obtained more successful results and continued using those. This is blind incremental trial and error not learnt or deduced information.
There's nothing revolutionary about this work in the slightest
PS If anybody has evidence of PRIOR ART, I hear god is trying to build a portfolio to refute the inevitable law-suits (course he's gona get screwed cos the devil has all the best lawyers
Flying is easy. Landing, well one bad landing can ruin your whole day....
You'd think that a robot like this would have an easy time flying. I mean, all you have to do is leave out the bit where you program it to know about falling, right?
Most exciting phrase in science: not "Eureka!" but "Hmm... That's funny..." -Asimov (abridged for \. limits)
Comment removed based on user account deletion
Nope, no sig
Here's how:
- Go find a decent AI or machine learning text such as Mitchell's Machine Learning or even Russell and Norvig's Artificial Intelligence
- Look up "unsupervised reinforcement learning"
- Buy some hardware, interface it to your computer. (This is the hard part.)
- Pick a suitable evaluation function, i.e. a fuction that evaluates the performance of a "run" and gives it a score. Perhaps something like "time averaged height of mechanism off the floor"
- Write code. Tweak slightly.
You could think of unsupervised reinforcement learning as "strategic thrashing" --- at first it doesn't know how its "controls" affect the quality of its performance, so it just tries different actions. Slowly it begins to learn that certain patterns of actions improve its evaluation function score and thus begins more directed "thrashing". If they wanted to do some real research, they could at least start with a mechanism that could fly. That would be impressive...when one is flying, isnt an error a crash?
People who think they know everything really piss off those of us that actually do.
...there's very little knowledge of genetic algorithms here, which is what this experiment sounds like.
It's the idea that complicated AI-like behavior can be "evolved" from a small set of simple abilities.
In a recent experiment, one researcher started downloading random configuration bit streams to an FPGA. They set up an evaluation metric to pick out the bits streams that were close to what they wanted, and interbred them via bit swapping.
Eventually they got a circuit which can tell the difference between the spoken words "Stop" and "Go". They "evolved" a circuit that implement speech rechognition with a fairly small amount of gates, and they have no idea how it works.
In fact, the final bit stream loaded into a different FPGA didn't work. The genetic approach actually took advantage of subtle differences (internal crosstalk, perhaps) between two chips of the same type.
This is all very interesting stuff. Maybe it's not "real" AI if your vision of AI is a gazillion byte monster coded by 10,000 people, but it does have one big advantage: it's useful. I believe some of the circuit synthesis tools out there are already using some genetic algorithms.
There's even evidence that our own brains, when you get in real close and real deep, are a huge collection of little system with relatively simple rules that self organize based on feedback in that precious time between birth and when the child can recognize the words "Stop" and "Go."
Sounds like a load of bull to me.
The feedback mechanism of the robot was a sensor that told it how 'high' it was getting off the ground. It would vary sequences of motor moves to try and maximize height.
BUT THE DAMN THINK CANT FLY BECAUSE ITS TOO HEAVY!
So how the hell could it come up with a flying motion as an 'optimal' solution? You'd think the thing would hop, or climb like it did, not flap. So, it my opinion it sounds fishy.
I think the robot simply had a laugh sensor and just made variety of different motions until the laughter of the audience reached a certain threshold. The Moon Walk was a runner up.
This article is interesting, however, because it moves the agents into the physical world where it isn't possible to obtain the same kind of idealized environments that are possible in silico.
Amazing magic tricks
When jerkychew takes over the world, he will be twitching and muttering, "I tried to warn them! I did! THey didn't listen..."
[o]_O
Another thing that got forgotten was the fact that against all probability a sperm whale had suddenly been called into existence several miles above the surface of an alien planet.
And since this is not a naturally tenable position for a whale, this poor innocent creature had very little time to come to terms with its identity as a whale before it then had to come to terms with not being a whale any more.
This is a complete record of its thoughts from the moment it began its life till the moment it ended it.
Ah!.. What's happening? it thought.
Er, excuse me, who am I?
Hello?
Why am I here? What's my purpose in life?
What do I mean by who am I?
Calm down, get a grip now... oh! this is an interesting sensation, what is it? It's a sort of... yawning, tingling sensation in my... my... well I suppose I'd better start finding names for things if I want to make any headway in what for the sake of what I shall call an argument I shall call the world, so let's call it my stomach.
Good. Ooooh, it's getting quite strong. And hey, what's about this whistling roaring sound going past what I'm suddenly going to call my head? Perhaps I can call that... wind! Is that a good name? It'll do... perhaps I can find a better name for it later when I've found out what it's for. It must be something very important because there certainly seems to be a hell of a lot of it. Hey! What's this thing? This... let's call it a tail - yeah, tail. Hey! I can can really thrash it about pretty good can't I? Wow! Wow! That feels great! Doesn't seem to achieve very much but I'll probably find out what it's for later on. Now - have I built up any coherent picture of things yet?
No.
Never mind, hey, this is really exciting, so much to find out about, so much to look forward to, I'm quite dizzy with anticipation...
Or is it the wind?
There really is a lot of that now isn't it?
And wow! Hey! What's this thing suddenly coming towards me very fast? Very very fast. So big and flat and round, it needs a big wide sounding name like... ow... ound... round... ground! That's it! That's a good name - ground!
I wonder if it will be friends with me?
And the rest, after a sudden wet thud, was silence.
Curiously enough, the only thing that went through the mind of the bowl of petunias as it fell was Oh no, not again. Many people have speculated that if we knew exactly why the bowl of petunias had thought that we would know a lot more about the nature of the universe than we do now.
"How Robots took over the world" circa - 2051.
Electronic edition, of course.
If your bitterest enemies are people who hack the heads off civilians, then I would say you're doing something right.
" A robot has taught itself the principles of flying -- learning in just three hours what evolution took millions of years to achieve, according to research by Swedish scientists published on Wednesday. "
WRONG! This creature was created/born with the neccesary equipment to fly. The evolution argument is for how long it took birds to grow wings, or whatever they believe. 3 hours is how long this "newborn" learned to fly, which is comprable to some other advanced birds.
And my second point: Evolution and Creationism are both theories that should not be stated as fact in any publication that wants to remain credible in the eyes of true scientists who use the scientific method before stating something as fact.
The scientists' next project will be developing a robot which learns how to sky-dive by trial and error.
What the research does do is to lend credence to the argument that insects and birds could have evolved, rather than having been 'designed' by some sort of a God.
And how drunk were you when you wrote this post? The robot was designed, wasn't it? By human hands [read: higher intellignece, "God" to the robot], yes? This actually lends credence to the fact that incects and birds were created and designed by God. This proves absolutly nothing for evolutionary arguments, rather it hinders the arguments because this robot required the design of a higher being, Mankind! Geesh, some people should not be allowed to post when they are drunk.
Come on... Learning how to flap your wings isn't the same thing as flying... you should change the name of the story to something "Robot learns how to flap its wings".
So does that mean if 1000 robots typed 1000 words every minute for 1000 years, they would come up with the complete works of Shakespeare? And if so, does that mean monkeys will learn to fly?! WAIT! THEY ALREADY KNOW HOW TO! *as seen in the Wizard of OZ*
I've taught a computer how to play tic-tac-toe and checkers using basically the same technique. It's very entertaining actually. 1. Give it rules, IE you cannot move off of the board, you must only move once per turn, etc. 2. Set up 2 seperate AI's to play against each other. 3. Give it winning conditions, so that it knows if the moves it picked let it win. 4. Give it cool graphics ala wargames. After letting it run overnight, come back and play checkers against it. Beat us pretty nicely (I suck at checkers).
Wow, not a single paranoid reference to the T-1000. You know they're going to turn on us and hunt us down.
Yes, I'm kidding
1.Equiping a test subject with wings short-circuits the most intreguing part of the experiment.
2.Equipping a winged test subject with a moter too heavy to maintain loft is stupidity at work.
3.Thrust is not lift. Flight requires both, but this was thrust. The robot recreated 19th and 20th century flying machines. They didn't work either.
4.Horizontal stabilizers (vertical rods) are not considered to have been available during the evolution of flight.
The test is intriguing, for sure. But to bill this as AI learned flight is either poor press coverage, or a scientist seeking funding through an uninformed press.
--- "1.21 Jigawatts!" -Doc
Some of the people who believe this is not a form of evolution might find this program helps to explain how it works. Seems like evolution to me. Just because in the article the body was pre-designed, doesn't mean the mind wasn't evolved.
I started a Cybernetics degree at Reading University about 7 years ago. At the time they had just been featured on TV demonstrating their amazing set of 7 robots (named after a certain set of Dwarfs immortalised by Disney) which SIMULATED the behaviour of sand wasps. Wow I mean they could simulate insect behaviours - incredible! This must have been a huge breakthrough right? Erm....no. In one of the lab experiments we had to re-program portions of their behaviour. The object avoidance code consisted of (it wasn't C, but this is the equivalent):
h tWheel); //Hunt for new target
:)
while(1)
{
MoveForward();
while(Sensors()==OBJECT_AHEAD)
TurnRight();
}
The code to "follow" another "sand wasp" (they all had beacons on the backs of them) was something like the following:
while(1)
{
bool bNeedToDriveLeftWheel=false;
bool bNeedToDriveRightWheel=false;
if(BeaconPickedUpByRightSensor())
bNeedToDriveLeftWheel=true;
if(BeaconPickedUpByLeftSensor())
bNeedToDriveRightWheel=true;
if(bNeedToDriveLeftWheel || bNeedToDriveRightWheel)
MoveSandFly(bNeedToDriveLeftWheel,bNeedToDriveRig
else
MoveSandFly(false,true);
}
Not what you would call rocket science I think you'd agree. But the head of cybernetics (Known on http://www.theregister.com as Captain Cyborg for his other idiotic media antics) managed to get on national TV with them. Hence I have a certain amount of cynicism when these people claim any kind of breakthrough.
Incidentally I also remember him getting on NewsNight with an "AI" program which was "deciding" if it wanted to eat a hamburger or not. My code for that:
if((rand()%2)==1)
EatBurger();
else
WatchWeight();
Jeremy Paxman quite obviously hated his guts, as he can sniff out a fraud a mile off
BTW I transferred my ass outa there after the first year onto the Comp-Sci course and never looked back!
This is a link to the research group in Sweden (I used took a few classes there).
/ ro botics.html
http://www.frt.fy.chalmers.se/cs/pages/robotics
- Tor
Then after being told its attempts will fail since they have failed every other numerous time it has tried them and predictably failing, it will blame it on ANYONE who is not extreme as it is and generalize them all in a category called 'conservative'. (as in vast right wing conspiracy) Ignoring that its own tactics and rhetoric is by definition antithesis to the stated goals and thus hypocritical and paradoxical, it will continue on its hateful way spreading gloom, distrust, malice and scorn instead of actually being a part of the SOLUTION to solving problems.
Although I guess if a liberal could have a 'logic' module installed they would cease being a liberal.
"Hey bob, what do you think about this issue"
"Well Al, I don't know as I have not yet heard from my masters what I should think about it"
"You mean you don't apply critical thought, consistent criteria and self evaluation along with logic, reason and true concern for the long term results?"
"hahaha, you must have just falled off the apple cart"
This robot didn't "evolve" as researchers (or the article) claim. The robot "learned", which is entirely different. Claiming it evolved at lightning speed, somehow implying that its computerized "evolution" is superior to that of living animals, is scientific masturbation.
Evolution is when you only have arms or legs, then over time (millions of years in terms of animal species) your appendages become wings that are capable of flight. An animal merely having legs or some form of primordial wings cannot learn to fly in three hours time, because it is not physically capable. And it cannot change its appendage into a wing no matter how hard it tries.
Learning is when you have a working wing, and through trial and error, or some instinctive a priori knowledge, figure out how to use your fully functioning wings to fly. Some birds do learn to fly quite quickly, once their wings have matured to the point of usefulness when they're young. That's what this robot did, to some extent.
While it does appear to be a very interesting experiment, in no sense did the robot evolve. It merely learned how to use what it was already given. Should it have been created with no wing, and subsequently created self-replicating robots that eventually did have usable wings, then perhaps one might say it evolved.
It's not really that hard to learn locomotion in a continuous environment, where you can improve a little bit at a time. Swimming and crawling were done years ago. Basically, you formulate the problem in terms of a measurement of how well you're doing, and provide a control algorithm with lots of scalar parameters that can be tweaked. You then apply a learning algorithm which tweaks them, trying to increase the success metric. Any of the algorithms for solving hill-climbing problems in moderately bumpy spaces (genetic algorithms, neural nets, simulated annealing, adaptive fuzzy control, etc.) will work.
There are limits to how far you can go with this technique, and they're fairly low. Gaits that require balance, such as biped walking and running, require more powerful techniques. Any task that requires even a small amount of prediction needs a different approach. Rod Brooks at MIT has explored the no-prediction no-model approach to control thoroughly, so we have a good idea now what its limits are.
I am trying to teach a robot to go, "Bok Bok Bwaaaahk Bok Bok."
If I succeed, do I get a slashdot story?
Table-ized A.I.
Something more acheivable would be a device which "flips the birdy" everytime somebody nearby honks.
when can me 'en jeb shoot her down?
Interesting article, but a misuse of the term Evolution. The machine didn't evolve, it learned. Evolution doesn't exist.
..to yesterday's item, "Scientist Crushed To Death By Falling Robot In Failed Flight Attempt"
~Philly
No no no... we're going about this all wrong. The key was in the random objects they left lying around. If one had been a suitcase the robot had lost years before on another planet, it would have been suitibly distracted to fly.
Remember: to fly, you should throw yourself at the ground -- and miss.
You are right about the GA, but this is closer to Genetic Programming than GA. GA evolves the 'answer', but GP evolves the 'solution' to the answer.
:)
There's a difference, GA is much easier to program than GP and is usually much faster. Example of a good candidate problem for to use GA to solve would be the travelling salesman problem, while GP would find a method to solve a problem.
While fundamentally pretty similar, GP is slightly more complicated. You have to deal with issues such as program over growth during evolutions, which doesn't happen in GA.
http://www.genetic-programming.org/ is a good source to learn more about GP.
I took a class on GA and GP while in college, very interesting stuff.
geek page at KY speaks
In biological evolution, the figure of merit is approximately reproductive success. Evolution works by favoring genes that get themselves copied a lot.
Exhaustive search would be the obvious way to find a true global maximum in this sort of problem, but often the cost-per-tuple of evaluating the fitness function is non-trivial. In cryptographic key searches, the fitness function is a Dirac impulse which is one at the correct key and zero everywhere else, so near-misses don't help you to find the global maximum. (This isn't strictly true; many crypto algorithms have classes of weak keys, but that's a diversion for another time.)
Another partial search algorithm, aside from GAs, is simulated annealing. Yet another is the backpropogation algorithm used to find good sets of weights for neural nets.
As the crypto example illustrates, these partial search algorithms rely on gradients of the fitness function, where near-misses have higher fitnesses than wild-ass misses. One of the things one does when using a GA as a design tool is therefore to try to select fitness functions with gently sloping gradients throughout the space of solution tuples.
WWJD for a Klondike Bar?
trial and error? if only intuition could be programmed!
I believe that the time is rapidly approaching in which we will be able to turn over ALL of our problems to machines. Then the human mind and its creativity and ingenuity can be used to discover the new problems and challenges while the machines can be left to solve them.
We seldom regret saying too little but often regret saying too much.
The posters who are saying "You stupid creationist idiots are proven wrong once again" are comletely missing the point of this article. Biological evolution is not even in question here -- this article is about evolutionary computing.
Any trait can be evolved, if it is defined specifically enough, the iterations that approach it are promoted effectively and the ones that detract from it are discouraged. The interesting thing about this story is that the programmers created a learning scenario for the computer/robot that was able to provide the right kind of stimulus and "culling" of the motions so that the desired result was achieved by simply defining the desired result and letting the program go off on its own.
Any sufficiently simple magic can be passed off as mere advanced technology.
You went way over my skill level in the first part of the first step. ;-)
Any sufficiently simple magic can be passed off as mere advanced technology.