Humanoid Robots Playing Soccer, Part 2: How They Work

In July, the US scored a major international soccer victory. While the nation was unable to bring home the Women’s World Cup, Team DARwIn took first place in the Humanoid Kid Size competition at the 2011 RoboCup tournament in Istanbul, Turkey. The robots, whose name stands for “Dynamic Anthropomorphic Robot with Intelligence,” are a collaboration between the University of Pennsylvania and Virginia Tech. Penn Engineering team members who traveled to the competition included Stephen McGill, Seung-Joon Yi, Yida Zhang, along with Jordan Brindza, Ashleigh Thomas, Spencer Lee, and Nicholas McGill, who are undergraduate and graduate students in the General Robotics, Automation, Sensing and Perception (GRASP) Laboratory. Team members from Virginia Tech included Jeakweon Han, a Ph.D. student in the RoMeLa Lab, and Taylor Pesek, an undergraduate in Mechanical Engineering For the competition, Penn developed the software framework that provided each robot with artificial intelligence (AI). This AI operates on multiple levels. At the most basic, it provides instructions on how to move each joint of the leg in order to walk; at the most complex, it incorporates all of information gathered by the robot—such as whether a collection of red pixels in its camera represents the ball, or the distance between itself, other robots and the goal — and uses it to make gameplay decisions. The engineers their team’s overarching strategy should focus on speed at the expense of strength and accuracy

Comments are closed.