Manuela Veloso: Fighter for Robofootball
This researcher in the field of robotics has turned robot football into a global phenomenon.
Leaving the elevator on the seventh floor of the Gates Computer Science Center at Carnegie Mellon University, I stumbled upon a clumsy but courteous robot who greeted me and escorted me to the office of Manuel Veloso , who, seeing her, shone as if proud of his child. Then she pressed a few buttons and sent the robot to her laboratory, which was further along the corridors.
')
Veloso, a computer science professor at Carnegie Mellon University (CMU) in Pittsburgh, has been working for more than two decades to create such autonomous mobile robots. She believes that the day will come when humans and robots will coexist, and my robot guide, named CoBot (which means Collaborative Robot, collaborating robot) is one example of her contribution to creating this future.
Its other, much more significant contribution is that it was the driving force that turned robot football into a global phenomenon. The robot football competition RoboCup , which it has supported since its birth, has not only greatly advanced robotics, but also has set many on this discipline, having formed of brilliant minds of skilled researchers and designers. The 2013th year competition in the Netherlands attracted 410 teams from 45 countries and 40 thousand spectators. It would not be an exaggeration to say that a robot football would not have taken off so swiftly if it were not for Veloso's pioneering research, her energetic support and accurate foresight.
Veloso introduced the use of artificial intelligence techniques to the RoboCup competition, and, according to Minoru Asada , a robotics professor from Osaka University, who is one of the founders of RoboCup, has fundamentally changed the game. As a leader in RoboCup, in related research topics and in pedagogy, Veloso greatly advanced robotics, says Professor Assad. He calls the work of Veloso "an invaluable bridge" connecting robotics and artificial intelligence.
Veloso goes swiftly, and speaks abruptly. It seems that her body is overflowing with barely contained energy. In the early 90s she was a freshly baked professor at Carnegie Mellon University, where she received her doctoral degree, and had interests in artificial intelligence. In 1994, at an AI conference, she saw a rough demo model of a football-playing robot. Despite the fact that she grew up in a football country like Portugal, she was not very interested in this sport (she preferred American football). But, according to her, as a task for autonomous systems, football was great. And so she became one of the first researchers to seize this topic.
Initially, she began developing computer-simulated football, with her student Peter Stone , now a computer science professor at the University of Texas at Austin. Soon after, Veloso, with her other students, began to design a group of small robots that could receive information about the environment, draw conclusions, act, interact, and self-study in the process of moving around the playing field. However, in the nineties to build a robot was not an easy task, and it took her team two years to get all the engines, sensors and controllers to work properly. An even more tricky task was to figure out how to make five such robots act together.
According to Veloso, his colleagues often skeptically remarked in those days, “What the hell are you trying to do with these five robots?” At that time, researchers worked on special-purpose single robots, designed such robots that could go to Mars or the mine dig the ore. “Everything revolved around the development of a lone robot that can perform complex tasks,” she says. But Veloso had fierce plans.
In 1997, Veloso became one of the founders of the Robofootball Federation RoboCup Federation and helped launch the first games. Subsequently, she held various leadership positions in the federation and still advises the team of her university. According to Veloso, quite a lot of changes have occurred with the game during its existence. If at the very beginning the researchers were happy, if their football machine gunner was moving as it was programmed, now teams consisting of humanoid robots play for real, using defensive and offensive tactics and scoring a penalty.
This complication of the game owes much to Veloso’s own contribution to AI. In the Small Size Robot League , where small-sized robots compete, the Veloso team was more likely to reach the finalists. The robots of her team, the size and appearance of a jar of coffee, plan their actions with incredible accuracy, and hit the ball with such force that he even left a few dents on the walls of her laboratory. The first robots the team performed with were assembled from Erector sets, but even then their intellectual abilities were superior to those of other machines. Veloso's secret weapon was a machine vision algorithm that used lines, patterns, and colors to help each robot track the ball and other robots.
In 2009, Veloso shared with the other participants of the competition the computer vision software developed by her team, using an open source approach that leveled out the capabilities of the participants and raised the bar for RoboCup competitions.
The official goal of the RoboCup competition is to create a team of robots that can beat the team of people of world champions of the year 2050. Indeed, the highlight of the game is the final match in which the best team of robots and the amateur team of people compete.
However, the task of beating people on the football field is actually a means, not a goal, says Veloso. The real goal is both the improvement of robots, and honing the skills of robotics. In the end, a robot that can accurately pass a pass can also serve breakfast. “The idea is to take the robots out of the playing field and put them into everyday life,” says Veloso. She played an important role in creating new categories of RoboCup @ Home , RoboCupRescue , and RoboCup @ Work competitions, which are designed to check robots in realistic scenarios for working at home, during emergencies and in production.
“Manuela is practicing a truly holistic approach to robotics,” says Stephan Zickler , a software engineer in robotics from iRobot , located in Bedford, Massachusetts. Manuela was his supervisor when he was working on his Ph.D dissertation. Instead of concentrating on one narrow area of ​​study, she always looks broadly at the whole picture, and encourages her students to do the same. Zikler argues that her broad outlook and absolute honesty makes students well-educated, intelligent researchers.
In his studies with autonomous machines, Veloso prefers efficiency rather than showiness. CoBot is an excellent example. He doesn’t look good: no hands, no legs, just a tablet, a camera, and a set of sensors, driving around on a chair on wheels. However, he has the ability to autonomous navigation in a changing environment, full of uncertainties, such as out-of-the-place chairs and half-open doors. Easy for humans, but puzzling for most robots.
When CoBot comes to a dead end, it uses a simple, but at the same time a brilliant strategy. He asks for help. When he encounters, say, the need to press the elevator button, he is looking for the nearest person to help him. Veloso calls this concept "Symbiotic autonomy."
Just as she foresaw that robot football will be a powerful tool that inspires a new generation of robotics, Veloso also sees great potential in such robots as CoBot and other thinking machines. One day they will help people in the office, in the workshop, and at home. This is a task over which work is underway, and perhaps it will go on forever. “A remarkable feature of AI and robotics is that this work is endless,” says Veloso.
about the author
Prachi Patel is a freelance writer and freelance editor for IEEE Spectrum magazine. He lives in Pittsburgh, USA.
An abridged version of this article was published in the print version of IEEE Spectrum magazine, entitled “Robot Football Champion” (RoboCup's Champion.)
Leaving the elevator on the seventh floor of the Gates Computer Science Center at Carnegie Mellon University, I stumbled upon a clumsy but courteous robot who greeted me and escorted me to the office of Manuel Veloso , who, seeing her, shone as if proud of his child. Then she pressed a few buttons and sent the robot to her laboratory, which was further along the corridors.
')
Veloso, a computer science professor at Carnegie Mellon University (CMU) in Pittsburgh, has been working for more than two decades to create such autonomous mobile robots. She believes that the day will come when humans and robots will coexist, and my robot guide, named CoBot (which means Collaborative Robot, collaborating robot) is one example of her contribution to creating this future.
Its other, much more significant contribution is that it was the driving force that turned robot football into a global phenomenon. The robot football competition RoboCup , which it has supported since its birth, has not only greatly advanced robotics, but also has set many on this discipline, having formed of brilliant minds of skilled researchers and designers. The 2013th year competition in the Netherlands attracted 410 teams from 45 countries and 40 thousand spectators. It would not be an exaggeration to say that a robot football would not have taken off so swiftly if it were not for Veloso's pioneering research, her energetic support and accurate foresight.
Veloso introduced the use of artificial intelligence techniques to the RoboCup competition, and, according to Minoru Asada , a robotics professor from Osaka University, who is one of the founders of RoboCup, has fundamentally changed the game. As a leader in RoboCup, in related research topics and in pedagogy, Veloso greatly advanced robotics, says Professor Assad. He calls the work of Veloso "an invaluable bridge" connecting robotics and artificial intelligence.
 | Manuela Veloso Honorary Member of IEEE Age: 57 Profession: Robotics researcher, Carnegie Mellon University Location: Pittsburgh Key Activities: Support international robots competition for training young researchers and improving autonomous robots. |
Veloso goes swiftly, and speaks abruptly. It seems that her body is overflowing with barely contained energy. In the early 90s she was a freshly baked professor at Carnegie Mellon University, where she received her doctoral degree, and had interests in artificial intelligence. In 1994, at an AI conference, she saw a rough demo model of a football-playing robot. Despite the fact that she grew up in a football country like Portugal, she was not very interested in this sport (she preferred American football). But, according to her, as a task for autonomous systems, football was great. And so she became one of the first researchers to seize this topic.
Initially, she began developing computer-simulated football, with her student Peter Stone , now a computer science professor at the University of Texas at Austin. Soon after, Veloso, with her other students, began to design a group of small robots that could receive information about the environment, draw conclusions, act, interact, and self-study in the process of moving around the playing field. However, in the nineties to build a robot was not an easy task, and it took her team two years to get all the engines, sensors and controllers to work properly. An even more tricky task was to figure out how to make five such robots act together.
According to Veloso, his colleagues often skeptically remarked in those days, “What the hell are you trying to do with these five robots?” At that time, researchers worked on special-purpose single robots, designed such robots that could go to Mars or the mine dig the ore. “Everything revolved around the development of a lone robot that can perform complex tasks,” she says. But Veloso had fierce plans.
In 1997, Veloso became one of the founders of the Robofootball Federation RoboCup Federation and helped launch the first games. Subsequently, she held various leadership positions in the federation and still advises the team of her university. According to Veloso, quite a lot of changes have occurred with the game during its existence. If at the very beginning the researchers were happy, if their football machine gunner was moving as it was programmed, now teams consisting of humanoid robots play for real, using defensive and offensive tactics and scoring a penalty.
This complication of the game owes much to Veloso’s own contribution to AI. In the Small Size Robot League , where small-sized robots compete, the Veloso team was more likely to reach the finalists. The robots of her team, the size and appearance of a jar of coffee, plan their actions with incredible accuracy, and hit the ball with such force that he even left a few dents on the walls of her laboratory. The first robots the team performed with were assembled from Erector sets, but even then their intellectual abilities were superior to those of other machines. Veloso's secret weapon was a machine vision algorithm that used lines, patterns, and colors to help each robot track the ball and other robots.
In 2009, Veloso shared with the other participants of the competition the computer vision software developed by her team, using an open source approach that leveled out the capabilities of the participants and raised the bar for RoboCup competitions.
The official goal of the RoboCup competition is to create a team of robots that can beat the team of people of world champions of the year 2050. Indeed, the highlight of the game is the final match in which the best team of robots and the amateur team of people compete.
However, the task of beating people on the football field is actually a means, not a goal, says Veloso. The real goal is both the improvement of robots, and honing the skills of robotics. In the end, a robot that can accurately pass a pass can also serve breakfast. “The idea is to take the robots out of the playing field and put them into everyday life,” says Veloso. She played an important role in creating new categories of RoboCup @ Home , RoboCupRescue , and RoboCup @ Work competitions, which are designed to check robots in realistic scenarios for working at home, during emergencies and in production.
“Manuela is practicing a truly holistic approach to robotics,” says Stephan Zickler , a software engineer in robotics from iRobot , located in Bedford, Massachusetts. Manuela was his supervisor when he was working on his Ph.D dissertation. Instead of concentrating on one narrow area of ​​study, she always looks broadly at the whole picture, and encourages her students to do the same. Zikler argues that her broad outlook and absolute honesty makes students well-educated, intelligent researchers.
In his studies with autonomous machines, Veloso prefers efficiency rather than showiness. CoBot is an excellent example. He doesn’t look good: no hands, no legs, just a tablet, a camera, and a set of sensors, driving around on a chair on wheels. However, he has the ability to autonomous navigation in a changing environment, full of uncertainties, such as out-of-the-place chairs and half-open doors. Easy for humans, but puzzling for most robots.
When CoBot comes to a dead end, it uses a simple, but at the same time a brilliant strategy. He asks for help. When he encounters, say, the need to press the elevator button, he is looking for the nearest person to help him. Veloso calls this concept "Symbiotic autonomy."
Just as she foresaw that robot football will be a powerful tool that inspires a new generation of robotics, Veloso also sees great potential in such robots as CoBot and other thinking machines. One day they will help people in the office, in the workshop, and at home. This is a task over which work is underway, and perhaps it will go on forever. “A remarkable feature of AI and robotics is that this work is endless,” says Veloso.
about the author
Prachi Patel is a freelance writer and freelance editor for IEEE Spectrum magazine. He lives in Pittsburgh, USA.
An abridged version of this article was published in the print version of IEEE Spectrum magazine, entitled “Robot Football Champion” (RoboCup's Champion.)
Source: https://habr.com/ru/post/368657/
All Articles