@Home

2007 Results: 2nd Place!

Segway in Robocup @Home 2007 Final

@Home Explained

RoboCup @Home is an international research initiative that aims "to foster the development of useful robotic applications that can assist humans in everyday life" (from the Robocup @Home Rulebook). In the 2007 competition, robots in a living room and kitchen environment had to complete four of six specified tasks: navigate to a commanded location (Navigate), manipulate one of three chosen objects (Manipulate), follow a human around the room (Follow Human), search for and locate previously seen objects (Lost & Found), differentiate previously seen and unseen humans (Who Is Who), and copy a human's block movement in a game-like setting (Copycat). All teams performed a freeform demonstration, during which they showed off their most impressive technical achievements, to a panel of other team leaders. Five finalists performed demonstrations for trustees of the Robocup organization. In this final round, we came in second place.

Our Robot

Our home assistant robot is built from a Segway RMP base. The Segway senses its world with a laser range finder, a stereo camera, and a webcam. A tablet PC, attached to the top of the platform, evaluates sense data and outputs motion commands.

2007 Results

UT Austin Villa finished 2nd at Robocup @Home in Atlanta!

Eleven teams from Europe, Australia, Asia, and North and South America fielded excellent, competitive robots.

Of the 6 fixed tasks, teams could participate in as many as 4. We chose to compete in 3. In the Follow Human and Lost & Found tasks, we performed well, completing all aspects that we attempted. The Who Is Who task was a real success. In it, the robot's task is to meet 4 people and subsequently identify them once they have been rearranged. For good measure, a stranger is thrown in. Our robot scored higher than any other in this task. Details of our person tracking method can be viewed here.

Segway at Robocup @Home 2007

(from Time Magazine)

In the Open Challenge round, the Segway robot performed a demonstration that combined our person facial recognition, shirt recognition (as a secondary classifier), and human following capabilities. A jury of our @Home peers ranked us 2nd place among the 11 teams. With this, our place in the Final was secure.

The Final consisted of an initial presentation, a robot demonstration, and a question/answer period, each about 5 minutes long. For the Final we fine-tuned the demonstration and added a couple of twists. The panel of Robocup trustees ranked us 2nd of the 5 teams in the Final.

Video can be found below. The quickest way to get a sense of our robot is to watch the demonstration first.

View the full results here.

The @Home team is Peter Stone, W. Bradley Knox, Juhyun Lee, Youngin Shin, and Mohan Sridharan.

Videos