The robots are starting to open their eyes.
Last week, Google announced it was partnering with the chip manufacturer Movidius to put machine learning in mobile devices. Machine learning is a branch of artificial intelligence, where computers can be programmed to recognize patterns in data based on previous sets of data. It’s what allows IBM’s Watson to diagnose cancer patients and come up with recipe ideas, and part of what Facebook uses to recognize your face in photos. But most of these computations require sturdy internet connections to large servers to be able to sift through all the data.
Movidius’s new chips, on the other hand, don’t require a lot of power to run. They can do complex computations right on the device they’re in, without help from servers. And this could one day bring AI to smartphones and tablets, and even robots and drones.
“Instead of us adapting to computers and having to learn their language, computers are becoming more and more intelligent in the sense that they adapt to us,” Blaise Aguera y Arcas, Google’s machine intelligence research head, said in a video released by the two companies. As part of the new partnership, Google will license Movidius’ chips, and help the company with its technology.
Google wasn’t immediately available to share what how it hopes to use Movidius’s chips. But considering smartphones are one of its few consumer products, there could soon be a Movidius-powered Android device that uses its camera not only to see and record, but to make sense of the world, just like humans do.
The announcement has some pretty wide-ranging implications. At this point, drones are essentially flying smartphones, and one of the biggest obstacles to drone-based home delivery is that they keep crashing into obstacles. Researchers around the world—as well as major companieslike Amazon—are working on low-power technology to enable a drone to “see” and react if something gets in its way. If Movidius’s chips prove as energy-efficient as claimed, drones may soon be able to make it to their destinations in one piece.
At last month’s Consumer Electronics Show, Lenovo introduced the first smartphone that could see the world, the product of a partnership with Google’s larger computer vision commercialization project, called Project Tango. This phone would be able to tell you exactly where something is in a supermarket by connecting to the supermarket’s database, or measure exactly how big an object is by eyeballing it. Still others are looking for ways to incorporate this low-power computer vision tech into robots. Perhaps soon, Movidius and Google’s partnership will lead to robots that do more than just vacuum the floor or roll into walls.