Road to the Robocup 2004 87
RuiFerreira writes "Artificial Intelligence and Robotics researchers meet in Portugal from 27th June to 5th July in the 8th Robocup Football World Championships. RoboCup is an international research and education initiative. Its goal is to foster artificial intelligence and robotics research by providing a standard problem where a wide range of technologies can be examined and integrated. The RoboCup Federation proposed the ultimate goal of the RoboCup Initiative to be stated as follows: 'By 2050, a team of fully autonomous humanoid robot soccer players shall win a soccer game, complying with the official FIFA rules, against the winner of the most recent World Cup of Human Soccer.' Robocup has an exciting programme including RoboCup Symposium, the RoboCup Soccer (humanoid, middle-size, small-size, 4-legged, simulation), the RoboCup Rescue (real and simulated robots) and the RoboCup Junior (dance, soccer and rescue) competitions. The robotic competitions will take place at Pavilion 4 of Lisbon Industry Fair located at the Parque das Nações, the site of the 1998 World Exposition (EXPO'98)."
Re:2050 way too soon (Score:4, Insightful)
Classical AI has made little progress at all. Cyc is really a formal machine that is the epitome of classical AI, and all its bizarre limitations - and it doesn't do hardly anything. Autonomous robotics on the other has made leaps and bounds. If you think things like the QRIO and Asimo are impressive, wait till you see their platforms are merged with the latest navigation and communication abilities currently found in labs. Part of the point of the robocup is to show off the latest in teamwork abilities.
2050 is a reasonable estimate and a decent goal. You don't need a robot that will debate the meaning of life and philosophy with you to play good soccer. I somehow doubt many of those soccer players could do that very well, either
Consider how it wasn't all that long ago that assertions were being continously made that machines could never beat a human at chess...
Re:robo rules 2004 (Score:1, Insightful)
Re:2050 way too soon (Score:5, Insightful)
Your "gut feeling" is more accurately described as the "common-sense intuitive linear" view of the rate of change, and it would be wrong.
If you'd look a little closer, you'd notice that each evolutionary advancement took exponentially less time. Exponential progress is a feature of ANY evolutionary system, including technology.
From the Law of Accelerating Returns [kurzweilai.net]:
Robotics is just one advancing tech we'll see on the shortening road to the Singularity.
--
Re:Sounds easy. (Score:2, Insightful)
Although the Swiss are renowned for their impartiality this [bbc.co.uk] kind of makes me wonder if sour grapes didn't have something to do with it.
Although I am from England i'm not always patriotic and although we lost to France and later to Portugal, I think the better playing by the other team (France) meant they deserved it.
However, although I think they deserved to win. I also believe the disallowed goal, was the wrong decision, which would have meant the England win. I would have also doubted any further progression after that match.
Maybe referees shouldn't be from qualifying nations.
Of course none of this would happen I suppose if there were robot referees (or if they actually asked to see action replays, or listened to the linesmen).
Re:Sounds easy. (Score:4, Insightful)
Let's just say the backbone of the Portuguese national team are FC Porto players: Nuno Valente, Ricardo Carvalho, Jorge Andrade (former player), Paulo Ferreira, Costinha, Maniche and Deco. Finish up the Champions' League winner team with players the likes of Cristiano Ronaldo and Figo, and only ignorance could claim Portugal needs referee work to win against any Euro'2004 team. We did stumble the first time, because a naive Scolari thought he could rely on former glory players like Couto. No longer. He's on the right track, and now we're poised to be European Champions.
Re:Sounds easy. (Score:4, Insightful)
Re:Do keep up. (Score:1, Insightful)
Seriously, lately FIFA has done a spectacular job in trying to murder football. The only other sport so successful in shooting itself on the foot with idiotic rules and regulations is F1.
When will they learn?
Re:2050 way too soon (Score:2, Insightful)
I think robotics is the way to go, and it is also much more challenging...
I guess that what i was trying to say is that, if you look carefully, create a small robot that acts smart and actively interacts with the environment, MAY very well be more difficult than create a robot that can discuss the meaning of life.
QRIO and Asimo are impressive but they are way below the "smartness" of cats and mouses,
which simply means there is a long way to go in autonomous robotics.
Classical AI will reach its limitations very soon,
Will that will bring a machine that can discuss the meaning of life ?
I don't know but i think we will find out in the next 20 years.
Re:2050 way too soon (Score:1, Insightful)
You wouldn't expect it to do much, any more than you would expect your Encyclopedia Brittanica to become self-aware and launch a campaign to eradicate the ugly bags of mostly water.
The "classical AI" guys have have little recent progress essentially because they succeeded too well. Most of the major areas (natural language processing, planning, etc.) all created projects that solved the problems put to them as successfully and "intelligently" as any human.
In doing so, they discovered that the real problem turned out to be not the "intelligent" bits, but the limitations of the knowledge on which that intelligence was operating. Limited knowledge, rather than limited ability of the algorithms resulted in limited functionality. And coding up the knowledge by hand turned out to be a long and intricate process.
As a result, most of the "classical AI" researcher shifted their focus a bit. You'll see their papers with results in machine learning, knowledge representation, and other topics about building and accessing a large database of information. Cyc is a poster child for "we need a bigger database". Lenat decided that if that was the problem, best roll up the sleeves and get to work, and just started piling up knowledge for a decade. (We put humans to school for a decade or two; why should the AI work right out of the box?) But Cyc itself wasn't designed to solve any particular problem, but rather just to store and inference about a lot of basic facts.
Other of the classical AI guys simply moved into subfields not quite related to "intelligence" in the classic sense, like low-level perception or motor feedback, problems that "classic AI" was never aimed at in the first place. These aren't new paradigms to supplant the earlier work with the "proper" approach. They are new areas to be explored so that the classic AI isn't just a isolated brain in a jar.
The other remaining research area is integrating all this stuff into a generalized intelligence. There's probably not one single paradigm to rule all there, any more than humans operate in only one manner.
Re:Riots (Score:3, Insightful)
Besides which, the riotting usually comes when a team loses.