"Passing is difficult, [but] receiving the ball is hard," says Veloso. "It's not just that the robot has to see the ball pass by, but has to intercept it. For a pass-receive to be effective, this is not a trivial thing."
Veloso also said that her team of programmers has developed "other tricks" that she wouldn't disclose prior to the upcoming competitions. But she hinted that a lot of the algorithms developed by her graduate students had to do with ironing out other teamwork issues -- including adapting the robots to use "plays" or specific formations.
"All the other teams [in previous years] used the same formations," says Veloso. But "CMDash05 [robots] will be able to use different formations against different teams."
But for all the increase in power and capabilities available to them, developers say there are still major hurdles to solving the problem of teamwork among robots.
For one, better artificial intelligence routines could become too powerful, causing robots to spend more time analyzing and communicating with each other rather than reacting to the situations around them.
"If we are … in some team environment like playing basketball, you jump without asking first because you probably understand you're closest to get the ball. You implicitly understand that your teammates won't waste their effort in duplication," says Veloso. "With robots we have to program that or they will spend too much time communicating, and miss the opportunity [to act]. How much should they negotiate, how much is [left to] opportunity … those are still things we need to figure out."
But figuring out those hurdles will mean more research and funding -- something that the United States isn't pursuing as aggressively as other countries, says Veloso.
"You have to realize that the amount of robot development [in other countries] is orders of magnitude higher than here in the U.S.," says Veloso. "They [Japanese organizers] are developing Robocup in Japan and they are expecting more than 200,000 people to come. This is like their [national] passion that we can only imagine."
And that, she says, is putting her teams -- and the United States overall -- at distinct disadvantages.
"In 2002, we won Robocup and we're first in the U.S. [RoboCup competitions]. But last year, we lost to the German team," says Veloso. "They had 40 people working on their robots -- 10 students per robot. I have seven. So everything we do is on a much smaller effort and scale."
Still, the good news is that popular and research interest in America is gaining traction. Automated vehicles such as the Predator unmanned spy plane and iRobot's Roomba vacuum cleaner have made their way into the minds and lexicon of the public.
And more colleges and companies are joining the robotics race. Spelman College, with funding from the Coca-Cola Company and NASA, for example, will be the first U.S. undergraduate research team comprised entirely of black women to enter the international RoboCup competition.
Such sparks bode well for increased awareness and attention to robotics research -- one of the primary goals of RoboCup.
Still, professional athletes like David Beckham and Freddie Adu won't have much to worry about for quite some time.
For now, even the best RoboCup players can't play a soccer game that meets regulations put forth by Fédération Internationale de Football Association or FIFA. Most RoboCup matches, for example, are just two 10-minute halves rather than the two 30-minute halves of a regulation game. And while the larger "humanoid league" of RoboCup competitors may resemble and act like human athletes, matches among "bipedal autonomous robots" are limited to a two-versus-two game rather than a full field of 11 players.
But researchers say the robots are catching up -- and well within the goal of being able to play against human opponents in a FIFA-regulated game by 2050.