Communications of the ACM
Robots Use Kinect to Understand Our World
Photo courtesy of Microsoft
Cornell University researchers are teaching robots to understand the context of their surroundings so that they can pick out individual objects in a room.
"We have developed an algorithm that learns to identify the objects in home and office scenes," says Cornell's Hema Koppula, who is conducting the research with Abhishek Anand. The key to the system is Microsoft's Kinect sensor, which works with the algorithm to recognize particular objects by studying images labeled with descriptive tags such as "wall," "floor," and "tabletop."
The researchers used 27 labels, 10 each for office and home scenes and seven that applied to both. The system also can take relative locations into account, such as the fact that computers are usually found on top of a table, not underneath it. The algorithm was able to achieve 84 percent recognition success for office locations and 74 percent recognition success for home locations.
"The next aim is to also include humans in the learning process, [with a robot] observing humans and being able to learn attributes of objects," Koppula says.
From New Scientist
View Full Article
No entries found