Oh those MIT kids, always trying to improve the quality of their robots.
The tech institute's latest enhancement to robot technology focuses on eyesight. Using a "statistical construct" known as the Bingham principle, MIT graduate student Jared Glover increased robots' ability to identify the orientation of familiar objects in cluttered scenes by 15 percent over the next best model, according to MIT's website.
A robot's improved ability to differentiate orientation, Glover said, will lead to improvements in object recognition. A lesser program may not be able to identify a coffee pot at an unfamiliar angle, Glover's formula would stand a better chance. His research showed that his algorithm led to more correct identifications than the next best algorithm, MIT.edu reported.
Glover's work distinguishes itself by its flexibility in a field restricted by specificity, according to MIT's article
"You can spend your whole PhD programming a robot to find tables and chairs and cups and things like that, but there aren't really a lot of general-purpose tools," Glover said. "With bigger problems, like estimating relationships between objects and their attributes and dealing with things that are somewhat ambiguous, we're really not anywhere near where we need to be. And until we can do that, I really think that robots are going to be very limited."
According to Gary Bradski, vice president of computer vision and machine learning at Magic Leap and president and CEO of OpenCV, Glover's findings will be the building block for future work in robotics.
"It's a better representation, so I think once it's understood, this'll just kind of become one of the things that is built in when you're doing the 3-D fits," Bradski said. "[Glover] found something that was obscure, but once people are familiar with it, it will just be a no-brainer."
Glover hopes to add to his application of Bingham principles by teaching a robot to play ping pong, according MIT.edu.
