Better Mechanical Eyes Could Let Robots Play Pingpong
Oct 08, 2013 12:07 PM EDT
Researchers are developing a better eye for robots that could tell which way the handle of a mug is facing, and whether it was upside-down or not. This type of technology could help robots better-manipulate their environment.
In order to make this possible Jared Glover looked to a statistical construct called the Bingham distribution, an MIT press release reported. The researchers created a new robot algorithm based off the Bingham distribution that is 15 percent better at identifying surrounding objects.
The team was able to use the algorithm to allow the robots to analyze a pingpong ball moving through the air.
"Alignment is key to many problems in robotics, from object-detection and tracking to mapping," Glover said. "And ambiguity is really the central challenge to getting good alignments in highly cluttered scenes, like inside a refrigerator or in a drawer. That's why the Bingham distribution seems to be a useful tool, because it allows the algorithm to get more information out of each ambiguous, local feature."
The Bingham distribution is helpful because it provides a way to "combine information from different sources. Generally, determining an object's orientation entails trying to superimpose a geometric model of the object over visual data captured by a camera." Glover's robotic eye uses a Microsoft Kinect camera.
"For simplicity's sake, imagine that the object is a tetrahedron, and the geometric model consists of four points marking the tetrahedron's four corners. Imagine, too, that software has identified four locations in an image where color or depth values change abruptly - likely to be the corners of an object. Is it a tetrahedron?" The press release stated.
The researchers needed to figure out a way to take two of those points (the model and object) and seeing if they can be superimposed upon each other. Glover hopes to take a shot at aligning the points.
In studies, the robotic eye was able to correctly identify an object 84 percent of the time, which is one percent higher than the algorithm's competition.