Researchers at the Massachusetts Institute of Technology (MIT) including two Indians have developed a low-power chip that could help visually-impaired people navigate their environments. This super chip is much smaller and consumes a thousand times less power than computer chips, but is capable of performing equally complex computations.
About the size of a binoculars case and similarly worn around the neck, the system uses an experimental 3-D camera and a Braille interface. The user carries a mechanical Braille interface, which conveys information about the distance to the nearest obstacle in the direction the user is moving. Actual images of the device have not yet been released.
This device trumps all previous initiatives because they were too bulky. “There was some prior work on this type of system, but the problem was that the systems were too bulky, because they require tons of different processing,” says Dongsuk Jeon, a researcher at MIT’s Microsystems Research Laboratories (MTL), who now joined the faculty of Seoul National University in South Korea.
“We wanted to miniaturize this system and realized that it is critical to make a very tiny chip that saves power but still provides enough computational power,” Jeon added.
The algorithm maps the surfaces by taking random points and comparing them with neighboring points. If they lie on the same plane, the area is treated as the same surface. Accessing the chip’s memory is the most power-consuming operation. In order to minimize it, the team of researchers has configured the algorithm to always start at the upper-left corner of the point cloud, scans along the top row analyzing each point and comparing it only with the point on the left. When done with the top row, it moves to the points on the second row, but compares them only with the ones on the left and directly above.
Another functionality of the chip is that it compares each frame shot with the preceding frame. If very less number of points varies, the device determines the wearer is at rest. During this time, the 3D camera is instructed to capture on a low frame rate, thereby saving power.
The new chip and the prototype navigation system were reported in a paper presented at the International Solid-State Circuits Conference held from January 31 to February 4 in San Francisco.