When Robots Play Games 184
Roland Piquepaille writes "If the theory of evolution has worked well for us -- even if this is arguable these days -- why not apply it to mobile robots?, asks Technology Research News. Several U.S. researchers just did that and trained neural networks to play the Capture the flag game. Once the neural networks were good enough at the game, they transferred them to the robots' onboard computers. These teams of mobile robots, named EvBots (for Evolution Robots), were then also able to play the game successfully. This method could be used to build environment-aware autonomous robots able to clear a minefield or find heat sources in a collapsed building within 3 to 6 years. But the researchers want to build controllers for robots that adapt to completely unknown environments. And this will not happen before 10 or maybe 50 years. You'll find more details and references in this overview, including a picture of EvBots trying to find their way during a game." Read on for a similar robot competition held this weekend in France.
saunabad writes "The annual Eurobot autonomous robot contest for amateurs is held this weekend on La Férte-Bernard, France. This year's theme is 'coconut rugby,' and the robots are collecting small stress balls from the field and carrying them to the opponent's end, or shooting them in the rugby goal, while avoiding the randomly placed obstacles at the same time. Each team has a one main robot and an optional small assisting robot."
Landmines? I don't think that's quite necessary... (Score:5, Interesting)
What would be really cool? (Score:1, Interesting)
Some examples of the tasks a robot could do are judge criminal cases, mark exam papers, and moderate slashdot posts.
However although the robot will probably make the right choice more times than a human we still wouldn't trust these important decisions to them.
Re:Caution (Score:2, Interesting)
The day that these robots can play capture the flag the way I used to play it as a kid, I will bow to the robots and call them my master. Wading through water, climbing trees, and jumping through thick gorse were all commonplace whilst clutching the opponents' frisbee (for flags were hard to come by).
When the robots can climb that oak to retrieve the frisbee that was skilfully thrown up at the start of the game, I think it's fair to say that the robots may just beat us at capture the flag! It's a game, not world domination.
Re:The world needs a collaboration between (Score:1, Interesting)
Re here comes the "Bolo, Mark I" "Obsequious" (Score:2, Interesting)
They already have a backpackable mobile remote "eyeballs" robot that can roam building interiors while sending back pictures and other sensory data to the soldiers outside. Its not far from there to have a semi-autonomous small caliber weapon carrying robot which has been combat trained the same way these capture the flag bots were.
After many generations, once the training is complete the "State of mind" of the most successfull 'bots can be duplicated and copied into as many "x-thousand" of the little buggers as you want.
There is, of course, the small detail of solving the IFF (Interogate, Friend or Foe?) issue.
And how would the robot know when an enemy wanted to surrender to it ?
(Just a leetle closer lil' fella - I won't hurt you, I just want to surr (CrunCH!)... oops. excuse me. Did I step on you? )
Friday's Softer World strip = so on topic (Score:3, Interesting)
Read the rest from their homepage [asofterworld.com].
Re: How long for an evolutionary cycle? (Score:3, Interesting)
> What I wonder most is how long did every evolutionary cycle take?
When you're using real robots, it takes a fair amount of time since the robots actually have to do enough stuff to be measured for fitness.
OTOH, I've seen video games that could be evolved with video turned off, allowing entire games to be played in a fraction of a second. So some people are trying to get a rough solution by evolving in an accurate simulator, and then fine-tune the solution by additional runs on the robot after the simulator training.
> It would be quite nice that a robot could adopt himself to a new environment in let's say, 2 minutes.
I've seen a demo of this in a computer game, though not in a robot.
Look for more of this kind of stuff in computer games within just a few years.
Re:"even if this is arguable these days"? (Score:2, Interesting)
This disagreement is a direct result of the problem of having multiple meanings for evolution. Evolution can mean whatever the arguer likes - such that proof of one form of 'evolution' is then flipped to imply that all uses of the word 'evolution' have proof, which isn't the case.
Natural Selection and adaptation is, in effect, a reduction in the diversity of the gene pool. Harmful genes are removed, while beneficial ones are given freedom to dominate. This results in the loss of genes, but is called 'evolution'.
Then we also have mutations. These produce changes in an organism, but are almost always harmless/harmful. This process of introducing new diversity (however limited it may be) is also called 'evolution'.
A big problem of this debate is the fluid meanings of 'evolution'. That is why I tend to use the word "darwinist" instead of "evolutionist". The former describes a specific viewpoint on the origin of life, while the latter, scientifically described, would be:
"A change in allele frequencies in a population over time". With that I have no dispute. When the darwinists add more, then I have an issue.
Once upon a time on slashdot, simply saying anything against evolution was a surefire way of having you dropped to -1, off topic, flamebait, or troll, even if you were none of those. People would use their mod points to win an argument instead of responding. Nowadays the situation is much more tolerable - and some are coming to recognise that darwinism was adopted, not because of any amount of proof, but because it was popular. It has never enjoyed good evidence. That is why scientists are so eager to trumpet every latest discovery as proof of evolution. Unfortunately for them, the evolution that does have proof is the scientific one I gave above - changes in allele frequencies. That all living things share a common ancestor is unproven, and seems unlikely.
Re:Arguable? (Score:3, Interesting)
Anyone who has credible evidence will not be ignored, no matter which cult they do or don't belong to. If they don't have credible evidence to back up their claims, they will be ignored (or laughed at). It's as simple as that.
"Or is it completely unreasonable for one to outline the fundamental problems with the darwinist model?"
Please go right ahead, I'd really like to hear it.
Actually, given a statement like that, I insist that you back up your claims.
Re:Robots coding and coding robots (Score:3, Interesting)
I recall seeing some comments on this topic from researchers using genetic algorithms to evolve circuits. The evolved circuits worked really well, but nobody could get a grip on how they actually worked (at least, that was the case as of the time the article was written). A drawback of having code or circuits that are human-incomprehensible is that you don't know how they will respond to unexpected external inputs (of course, that's often the case with human-written, human-incomprehensible code as well).
I agree that it's really interesting. I expect there's a lot of good research to be done in this area, and we'll probably see some really cool stuff come out of it.