This page contains the following sections:

• Recent news concerning RoboCup 2009 in Graz, presented as articles,
• a brief introduction of RoboCup,
• a comprehensive list of rankings and game results of the humanoid league, and
• common RoboCup Informations concerning different leagues.

Just click on a news headline or the link “Read more” to get more information and regarding picture galleries. A comprehensive gallery of all article images is presented here.

In earlier times, the game of Chess was recognized as standard problem of articial intelligence. A computer that was able to defeat a grandmaster in a game of Chess would be assumed to show intelligence. Today, there are computers capable of such behavior, but none of them is recognized to be intelligent: Computers just win using superior calculating power that allows to calculate victory probabilities for entire branches of the very large game situation tree. They do not use human approaches to win.

Thus, the game of Chess is not any more a leveling board for determining whether computers are intelligent or not, because it takes place in a discrete and sterile world to which standard search algorithms can be applied. As a just as sterile, but significantly more complex board game, Go could raise the bar: It produces game situation trees of overwhelming size that can't be handled by computers in conceivable time.

However, just to change from Chess to Go would not satisfy all the criteria of a standard AI problem. Moreover, recent research results indicate the emercence of intelligent behavior to need situations: Intelligence seems to need a body percieving incomplete sensor data from a dynamic environment it is situated in. According to this sensor data, adequate actions need to be performed. This is why the game of soccer was chosen and the RoboCup competitions came to life. By pursuing research around the RoboCup, one hopes to elicit new expertise in sensorics, actorics, computervision, cognitive control and related fields. Last but not least, by creating perception mechanisms that shall match the human ones by 2050, one hopes to learn about human perception through bottom-up research. As an objective measurement bar, RoboCup teams m eet for tournaments. This year, the tournaments are located in Graz ¹⁾.

To avoid misunderstandings, I most explicitely state that I travel to RoboCup 2009 as part of the team NimbRo, Autonomous Intelligent Systems Group, University of Bonn. This will not influence the neutrality of my coverage, especially due to the fact that that all unmarked coverage is personal work. That's why it is published at my personal web site. However, the pure amount of information may be biased towards the RoboCup leagues NimbRo participates in.

Please see here to get more information about the participating teams.

The Kid-Size League goes through several phases, that are planned to be sufficient for 24 competing teams at maximum. Additionally, there will be a technical challenge on Friday, July 3rd.

In this phase, the crowd of 22 teams and two empty slots is split into 8 groups A to H. The 8 teams that performed best at the last year's competition are equally distributed among these groups. Newcomers are equally distributed as well, if possible. After the first round robin phase, the winner of each group directly proceeds to the second round robin phase, the second and third of each have to compete in the intermediate round.

The Intermediate Round consists of eight knock-out-games each performed by a second and third placed team of the first round robin phase. The winner of each game proceeds to the second round robin phase, the loser drop out of the tournament.

In this phase, four Groups I – L that consist of four teams (two intermediate round winners and two winners of first round robin) each play round robin. The first and second placed teams of each group proceed to the final round, the rest drops out of the tournament.

The final round starts with the quarter finals and ends with the final game. Additionally, a game for the 3rd place is performed.

What the soccer games are for the kid-size robots, the dribble and kick competitions are for the teen size. In addition, the teen-size competitions also feature technical challenges.

A dribble and kick challenge is somehow a mix between dribbling and a penalty kick: It is played 1 vs 1 robot, an attacker against a defender. The attaccing robot has to dribble the ball and kick a goal, the defender as a goalie has to prevent the attacker from doing so. Every robot attacks and defends five times, which makes a total of 10 games per two competing teams.

There are 8 teams split into 2 groups.

In RoboCup, there exist different leagues, kinds of robots, and research goals. Some of them are briefly introduced at this place. All leagues have in common, that the regarding robots perform the tournament games in an autonomous way. Human interaction with games is strictly prohibited (except for the case of defects, etc.).

In the humanoid league, robots with human-similar body perform tournaments, meaning two arms, two legs, abdomen and head that suffice certain proportions to each other.

Kid-size robots are of a size between 30 and 60cm and perform soccer games in teams of three robots. The rules of these games are adapted by those of the FIFA. In addition to the games, technical challenges are performed.

Teen-size robots are larger than kid-size ones (between 100 and 160cm). They do not play real games, but only perform technical challenges, like penalty kicks.

Different to teen- and kid-size, robots of the standard platform league are not selfmade by the teams, but bought, which narrows the competition down to a pure software competition.

Robots competing in the mid-size league drive on wheels and feature omnidirectional vision, which enables a faster and more dynamic game than in the humanoid league. However, the disadvantage of this league is the missing natural motivation.

In this relative new league, assignments for household robots are defined that allow for competitions. An assignment for a robot may be, for example, fetching something to dring, if a human guest requests it in a verbal way.

A further new league is the rescue league, which seeks for solutions in autonomous finding of human beings in disaster aeras. For testing purposes, there are heat blankets and CO2 Emitters hidden in the rescue area that need to be detected by the robots.