The Dutch team only managed fifth place at the European Championship for soccer robots, which was held in Amsterdam last weekend. After a chess computer’s victory over world champion Gary Kasparov, teaching robots to play soccer is a new challenge for computer scientists.
/strong>
RoboCup 2000
The Dutch soccer robots looked like vacuum cleaners on wheels. Every robot views the world through a wide-angle camera lens and recognises colours, sizes and shapes. By looking at the lines on the field, the robot knows its location; and when an opponent becomes larger, the robot moves closer, and vice versa.
On the playing field, measuring six by twelve meters, everything has its own colour. For example, one goal is blue and the other yellow, and if the robot sees an orange circle, it recognises it as the ball. “During a demonstration in Delft, a cameraman was lying at the edge of the field. The red, round head of the man looked like a football to the robot,” says Wouter Bokhove, a student in Technical Physics at TU Delft. “The robot wanted to kick the cameraman, who managed to get away just in time.”
A kick by a robot can be hard. Its feet, a kind of kick spoon, are driven by a compressed-air apparatus. The pressure is controlled electronically, making it possible for the onboard computer to adjust the speed of its kick. A robot soccer game consists of two ten-minute halves. During halftime, the storage battery and air reservoir of the kick mechanism must be replaced.
War zone
A robot team is comprised of four members who communicate via a technique called wireless internet communication. “They must interpret the situation on the field and try to score together,” explains the team captain, Dr. Pieter Jonker, of TU Delft’s Physics department.
Because the goalie must be the quickest at recognising the ball and the movements of opposing robots, it has a better onboard computer and two extra cameras, which enables it to look to the left and the right.
The Dutch team members are from the universities of Delft, Utrecht and Amsterdam. The project team Autonomous Interacting Robots (AIR) wants to build independent, self-learning robots. TU Delft is trying to improve the pattern recognition and kicking mechanism; Utrecht is striving to improve the robot’s strategic thinking; and in Amsterdam, university scientists are working on the computer codes.
Eventually, the engineers want to develop rescue robots that can be used for fire fighting in tunnels or for digging up mines in former war-zones. “If robots are to work during disasters, it’s important that they operate independently and co-operate with other robots without human help,” says Dr. Jonker. But first, they have to learn how to play soccer. The competition on the playing field stimulates the development of new techniques. The scientists hope that robots will be able to play against the human world soccer champions in 2050. (RW/MS)
See the Dutch story on robots ‘Voetbalrobots strijden om RoboCup in Amsterdam’ for a picture of the match.
The Dutch team only managed fifth place at the European Championship for soccer robots, which was held in Amsterdam last weekend. After a chess computer’s victory over world champion Gary Kasparov, teaching robots to play soccer is a new challenge for computer scientists.
RoboCup 2000
The Dutch soccer robots looked like vacuum cleaners on wheels. Every robot views the world through a wide-angle camera lens and recognises colours, sizes and shapes. By looking at the lines on the field, the robot knows its location; and when an opponent becomes larger, the robot moves closer, and vice versa.
On the playing field, measuring six by twelve meters, everything has its own colour. For example, one goal is blue and the other yellow, and if the robot sees an orange circle, it recognises it as the ball. “During a demonstration in Delft, a cameraman was lying at the edge of the field. The red, round head of the man looked like a football to the robot,” says Wouter Bokhove, a student in Technical Physics at TU Delft. “The robot wanted to kick the cameraman, who managed to get away just in time.”
A kick by a robot can be hard. Its feet, a kind of kick spoon, are driven by a compressed-air apparatus. The pressure is controlled electronically, making it possible for the onboard computer to adjust the speed of its kick. A robot soccer game consists of two ten-minute halves. During halftime, the storage battery and air reservoir of the kick mechanism must be replaced.
War zone
A robot team is comprised of four members who communicate via a technique called wireless internet communication. “They must interpret the situation on the field and try to score together,” explains the team captain, Dr. Pieter Jonker, of TU Delft’s Physics department.
Because the goalie must be the quickest at recognising the ball and the movements of opposing robots, it has a better onboard computer and two extra cameras, which enables it to look to the left and the right.
The Dutch team members are from the universities of Delft, Utrecht and Amsterdam. The project team Autonomous Interacting Robots (AIR) wants to build independent, self-learning robots. TU Delft is trying to improve the pattern recognition and kicking mechanism; Utrecht is striving to improve the robot’s strategic thinking; and in Amsterdam, university scientists are working on the computer codes.
Eventually, the engineers want to develop rescue robots that can be used for fire fighting in tunnels or for digging up mines in former war-zones. “If robots are to work during disasters, it’s important that they operate independently and co-operate with other robots without human help,” says Dr. Jonker. But first, they have to learn how to play soccer. The competition on the playing field stimulates the development of new techniques. The scientists hope that robots will be able to play against the human world soccer champions in 2050. (RW/MS)
See the Dutch story on robots ‘Voetbalrobots strijden om RoboCup in Amsterdam’ for a picture of the match.
Comments are closed.