Slashdot Mirror

They should switch to...robot players

The world cup of robot football was just held in Montreal, and has been going for more than 20 years.

I assist at a local high school with their Robocup program (admittedly we mostly use LEGO Mindstorms, but some of the senior students use Arduino) and I think that is an awesome way in.
Robocup
Get yourself whatever kit you like and try to make a soccer robot, or a rescue robot, it will give you a nice and clearly defined target to aim for, but will also allow you to experience the little technical challanges that you will need to overcome. Even if you have no intention of competing, it is a good base to start from, and if you have the time or the desire you can assist schools and kids getting involved in the hobby too.

There's also RoboCup: autonomous robots playing soccer.

What was that about running around a field and kicking a ball? http://www.robocup.org/

http://www.robocup.org/

- has been going for a very long time, and often involves more sophisticated physical robots.

Also, as many people have pointed out, this is not really all that original. One of the oldest Genetic Programming's PhDs (in 1980) evolved card players. Tierra pre-dates this stuff. Nothing surprising has emerged here, but I guess it is a nice popular science article for those not familiar with the field.

As others have pointed out, Roger Alsing's Mona Lisa is very cool, as is the work that went into evolving an aerial design for NASA:

http://ti.arc.nasa.gov/projects/esg/research/antenna.htm ... there are over 6000 papers published on Genetic Programming and a huge chunk of them are applications like this. EC techniques like this are just heuristic search algorithms for program spaces.

RS.

The field of AI is not comprised of a majority of researchers frantically trying to build an expert system that can pass a Turing Test. Visual data is complicated and building a system that can take that information and make use it in a very simplistic manner is non-trivial. Read some of scientific papers published by the authors of Acquine, and you'll see that their methodology (image processing, regression, Bayes' classification, decision trees, support vector machines, classification and regression trees, to name but a few) is anything but trivial.

Not only did they build something novel, but they built a system that does a good job of approximating human response to good/bad photography.

If you want to contest the true novelty of their work, through an academically-inspired claim that they combined existing technologies in a way that isn't terribly novel, rather than creating their own technology, then that's fine. However, the blanket statement that some researchers are trying to do "real work" and that Acquine isn't real work, is a giant red-flag indicating that you likely haven't got the slightest clue what actually goes on in the field of AI. Typically researchers like to tackle problems where the utility of their solutions isn't immediately obvious, the previous link to the RoboCup competition is a perfect example; who cares if you can build a robot that can play soccer? By your reasoning, that would be an incredible waste of time. Except, it's becoming the standard problem for multi-robotic systems research, and a large number of AI researchers are devoting significant time towards building RoboCup teams.

Why?

Simple, pick a real world task for a group of robots that "matters". Now decompose that task into all the subproblems that you would need to solve in order to have a robot complete the main task. Chances are, you're going to run into problems involving self-localization, team-work/cooperation, vision, data fusion, etc... All of those subproblems are being worked on and solved in different ways by researchers in the RoboCup challenge. And chances are, if you choose the methodologies used by the teams that win games you're likely to have chosen the most effective methodologies available in the field.

The true value of research isn't the end-product of each individual research-project. It's the end-product of many "useless" research-projects combined.

I guess the subject says it all. These are just fancy RC cars. I'd be about a million times more interested in seeing the FIRA RoboWorld Cup or RoboCup televised.

Why don't they just enter Robocup?

it'll probably one of this guy's robots

If you like to know what Robocup is about. Watch this wmv video of the highlights of Robocup 2005 in Osaka.

This versatile model could one day be used for automobile driver's assistance, visual search engines, biomedical imaging analysis, or robots with realistic vision

• There already exist working automated driving software. The december 2006 issue of IEEE Computers magazing was on them last month. Read about the car that drove a thousand miles on Italy's road thanks to Linux, no less.
• Visual search engine exist, at the research level. The whole field is called "Content Based Retrieval", and the main issue is not so much to search, but to formulate the question.
• Biomedical image analysis has been going strong for decades and is used every day in your local hospital. Ask your doctor !
• Robotic vision is pretty much as old as computers themselves. There are even fun robot competitions like robocup.

Straight computer science? Well, you'll probably be just a code monkey.

I'm not sure where you guys all went to school, but it sure wasn't where I'm at.

All the comp sci guys that I went to school with (including meself eh) ended up being roughly equal parts code monkey and mathematician. Not as good at coding as a software engineer (though some were better...) and not as good at math as a pure math geek (though, again, some were better...)

Is my university that different, or are people just going by casual observation? I mean, I'm working on my M.Sc. in Comp Sci, and so far it's mostly math that I've been doing... I'm only just starting to build a simulation of what I sketched out mathematically so that I can get some data to analyze, but still, I'm at around 75% research and math, and 25% coding.

For anyone who cares, I'm trying to find out whether or not the premises behind the design choices of Probabilistic Incremental Program Evolution (PIPE, by Salustowicz and Schmidhuber) hold true in the RoboCup domain.

Oh cool... can you hook a brutha up? I just submitted a grant application to get the gear my university needs to get into the small league competition ... I could use all the money I can get :)

I work with my university's RoboCuphttp://www.robocup.org/ team. We've been looking at Microsoft Robotics Studio http://msdn.microsoft.com/robotics/learn/default.a spx for doing a lot of our simulation (since it incorporates a plethora of features, and a great physics engine, and because Microsoft give us lots of *bling*).

The thing that struck me most while going over the studio was it's great tie-ins to real Lego robots, namely the old Mindstorms and the new NXT. It's pretty exciting and a great way for people to get into serious robotics without all the hassle of machining parts, soldering electronics, and programming firmware. Definitely a cool product.

Well, I can't agree with "poker is driving artificial intelligence". This obviously neglects a game which has no chance system, only works with multiple agents, allot of uncertainty and has little or no predictable variables: soccer!
You don't know how the wind will blow, you need to play together, change stratagies (like attack and defend) and ultimately predict the future of an uncertain system without grinding to a hold. Now that's a game that drives AI research.

• Simulation league
  There is some low-bandwidth communication allowed. (e.g. via shouting a limited amount of messages / minute)
  
• Small-size league
  Can communicate wirelessly, also with the team's external server. AFAIK, the computing power is usually located on that server instead of the robots.
  
• Middle-size league
  similar to small-size league
  
• Four-legged league
  The robots are autonomous, but can communicate with each other (through an AP). There is an external server, but it is only allowed to command "start" or "stop". It isn't allowed to tell the position of the ball or communicate strategies / announce passes / ...
  
• Humanoid League
  similar to four-legged league. The robots must still be able to play even if the communication stops.
  

...RoboCup2006 in Bremen, Germany

This is awesome. I saw heavily customised lego robots at Robocup four years ago. Now anyone with a minimal budget can join in on the fun.

Given the extra height you maybe when they try to negate a doorway. Perhaps a quick re-brand is needed?

Extreme robot-soccer boots RoboCup

That 'SIOX' object selection stuff looks really really cute; you have to wonder if it would come in useful for machine vision/AI as well.

SIOX was developed for echalk.
And echalk is developed by people of the AI-working group at the CS department of the FU Berlin.
And that working group is also behind the successful robot soccer team FU Fighters who are currently World Champion in the small-size league of Robocup and vice-champion in the middle-size league.

So it wouldn't surprise me if parts of it are already in use for "machine vision/AI"..

One of the divisions of the ongoing RoboCup competition (robotic soccer) is the Sony Four-Legged League, where researchers, including some pretty high-powered robotics groups from major universities, hack teams of Aibos to play soccer against other teams of Aibos.
RoboCup
Four-Legged League

In the video, the robotic hand is catching the ball with its fingertips, which to a human would be an incredibly difficult thing to do -- akin to catching a fly with chopsticks. A human would wait until the object being caught is somewhat within the opened hand before grasping and making the catch, which is probably more secure. However, I can certainly understand the need to catch the ball BEFORE it impacts the visual sensors in the palm of the robot hand.

Also, the idea of putting the sensors in the hand itself is nothing short of brilliant -- no more parallax correction to worry about. This is something our wetware is quite capable of dealing with, it's a survival function of just about every animal with eyes. Sometimes you just don't WANT your eyes and your hands (or paws) in the same plane. For one thing, eyes are delicate. For another, it's much easier to determine the distance to something if you've got a slight angle on it as you watch.

This makes me wonder -- is there another sensor outside the hand to assist in calculating the trajectory of the inbound ball? Or is there still a need to pre-select the ball size and speed?

Also, it seems to me that if you can make a robot catch a ball, you can also make a robot hit, kick, or punch a ball. These methods may lead to considerable improvements in "sporting" robots in general, such as those used in the Robocup. It was probably possible to build a golfbot before, since the ball isn't moving when you hit it, but this would broaden the potential applications to sports using a moving ball. I'd imagine such a robot would make an incredible table tennis player, since that game is all about reactions and doing something with the ball at the moment you hit it (impart spin, "deaden" the shot, take sharp angles) and doesn't require much running around. If said robot can keep the ball down, it's not going to be involved in the slamfests you see on ESPN2 with players running around 20 feet behind the table. That doesn't happen until someone makes the mistake of elevating the ball.

Mal-2

There are a surprising number of responses commenting on Sony dissaproving of Aibo hacking and whatnot.

Haven't you heard of Robocup?

http://www.robocup.org/

The goal of robocup: To build a team of robots that can play and beat humans at soccer (possibly without killing the human team)

My uni has a team competeing.

We do pretty well at the f180 and abio class comps.

A less snake-oil-ish approach would have been to set out more or less achievable tasks and start with those as competitions, e.g. dribbling, penalty kicking (without goalie, simply to understand the mechanics of ball control).

With the different leagues, I think they've already got very achievable tasks, it's just that they're running in parallel with the less achievable ones. For example, the simulation league and the small-size league do quite well, because people competing in those leagues mostly only need to worry about multi-agent coordination, with some simple visual perception in the small-size league. The 4-legged league introduces some more difficult tasks, with the humanoid league introducing even more difficult tasks.

An even better approach would have been to avoid the publicity seeking soccer competition altogether and use more wortwhile tasks such as search and rescue among the rubble or a robot server in a restaurant (as in fact they did in IJCAI 2003).

From the RoboCup page, it looks like this is now a more-or-less integral part of the RoboCup event. The set of challenges though is very different though: They're typically isn't any multi-agent aspect, the environment is much more chaotic, and there's typically a human remote-controlling the robot.

Now they want them to play soccer? This is flying cars all over again.

Psst... they already do play soccer:

http://www.robocup.org/02.html

I went to the International Robocup Challenge in Padua Italy 2 years ago and have to admit that it's just amazing what's being developed. While the human robotics competition side of things has only just started if anyone has seen the SDR (Sony Dream Robot) then you'll see that it'll probably take less than 45years to win. The Robocup challenge's whole goal is to produce a robot by 2050 that will beat the top human soccer team in the world cup. That goal has been around for quite a few years, and isn't exclusively for the Osaka robotics team. I became very good friends with one of the AI programmers of one of the Osaka robotics team, and amongst all the different human robots theirs was the best (the SDR wasn't competing, it was only dancing in the background). Don't forget their are more sections than just the Human robotics side. There is the Small robot league which has some of the fastest games of robotics soccer you'll ever see, there's also the Mid sized robots, which kick around real sized soccer balls and have been using the full 360deg viewing for some time. Rescue league has lots of potential, and coupled with the computer rescue simulation competition I can see in the future waves of small robots being controlled by a hot air balloon floating above a crash site as they scour the rubble for signs of survivors. The humanoid robots will be able to (hopefully) kick the soccer ball with such precision and accuracy that they just shouldn't miss. They will be able to run simulations to determine the best way to hit the ball. There is already a team of researchers trying to create mechanical muscles which will be faster and more efficient then human muscles but are powered by chemicals not straight electricity (solving the battery problem). Check out http://www.robocup.org/ or just email me for more info.

We've had a couple of these for the past year or so. They're currently being developed into soccer-playing robots as part of Robocup.

The goal there is to have a team of humans on segways play segway robots in an actual soccer match.

The project's website, which has a movie page with a ton of movies of the thing in action is here.

Mark Tilden, the designer of Robosapien (and of BEAM robotics for any of you who have heard of it) purposely left the Robosapien open for modification (god bless his soul ^_^) More people/companies should do this! =)

I'm actually quite interested in this, as I'm in the Australian Robocup Junior atm (Year 9) and this is a great platform to begin experimenting in bipedal robots. Heck, it could even begin it's own competition section (the robotic AIBO dogs have their own soccer competition)

Mark Tilden is also heavy researcher in the biomorphic robotics field, and through his experiments he has essentially created the BEAM robotics field.

Solarbotics, one of the best resources for BEAM robots I know of

Robocup Official Website

Mark Tilden at WIKI

is get together with the http://www.robocup.org/ people and some miniaturization experts, and come up with a total RoboFoosball table.
Forget spinning those sticks, you could coach the little guys! Against the inhuman RoboCoach!

Segway soccer is looking to become the newest Robocup league. Though it sounds a little silly, it is significant in that it will be the first Robocup league to feature competition between humans and robots, which brings Robocup a little closer to its goal of having a humanoid team that can beat the World Cup team by 2050.

--

Robotic Soccer's answer to the World Cup - it's called Robocup

Many universities take part in this event and the cup's mission is to "...develop a team of fully autonomous humanoid robots that can win against the human world soccer champion team by 2050..."

Say 46 more years, by the RoboCup organization's target.

I suppose it's all part of the ultimate goal of developing a team of fully autonomous humanoid robots that can win against the human world champion team in soccer

Robocup is the real world cup of robot soccer, about 5 years in the running. The 2004 American Open is happening in late April, and it will supposedly be open to the public. And for the really hardcore fans, there's Robocup 2004 in Portugal. See ya there ;) Failing that, you should at least watch a match (4-legged league).

I have seen this robot here in Padova (Italy) last July at Robocup competition. I also took some pictures as this little toy got my attention: check out!
Anyway a man was throwing some colored circles on the ground, and this brilliant robot was wandering around looking for them and once he could detect them he started walking from one to the other. Great!

You can imagine how hard it is to make a team of aibos play soccer autonomously!

The last games to be conquered by machines will be physical like soccer, or involve recognition of speech or visual information. Humans use their brains to their fullest in those activities; it's what they're best at compared to machines. When AI gets that far, if you believe in the Singularity, we'll be at it.

This would be a great place of employment for all those robot soccer players that will be out of work in America due to yet another instance of the American appetite for soccer coming in well below expectations.

"Bender like Beckham" indeed.

You mean something like Robocup mentioned previously on Slashdot?

I have no problems reaching the site. Actually, it loads quite fast, I dare to say. So, maybe the connection between the US and Japan is saturated :).

Universities are doing just that in the various RoboCup events.

The title of the press release is wrong. It is referencing the first RoboCup American Open which is happening this weekend.

You have already missed Neal Stephenson, but not the ASIMO demonstrations tomorrow.

hmm, well,

IMHO, this is the next "Turing Test" http://www.robocup.org/ . It's quite hard to play soccer. I tell you =)

It doesn't matter if they cannot play chess or do math.

Is intuition computable?

This year's competition is to be held in Padua, Itally July 2nd - July 11th.

These guys should hire some programmers (and engineers) from The Robomower Company... The mower can handle a quarter acre with all kinds of obstructions...

Also on my wishlist:
1. Return-to-base self-charging.
2. Return-to-base dust bin emtpying.
3. Environment learning. It could develop a map of the floor, and keep track of the dirt collected in different areas. Then it could do a daily cleaning of the high-traffic areas, and do occasional full passes.
4. Take some lessons from Robot Soccer and learn some teamwork. (Imagine a beowulf cluster of these!)
5. Remote Interface with an X10 burglar alarm. (Although "Release the vacuums!" just doesn't have the same ring as "release the hounds!)

--sg

Deeper RoboCup Link Heh, "And taking the field, it's the Sony Beowulf Clusters!"

When the robots become autonomous, battling without human intervention, then I'll start showing real interest.

You'd probably be interested in RoboCup Soccer. The goal of this ambitious robotics project is to develop a team of fully autonomous robots that can defeat the human champion team of soccer by the year 2050.

I would like the see the terrorists vs the RoboCup bots in fight. You can go in with info-red mounted cameras and hydraulic stabilized guns, but there would be nothing cooler than watching a ton of 5 foot robots slug it out with 'em UFC style.