Want CNET to notify you of price drops and the latest stories?

Wave fingers, make faces: The future of computing at Intel

The chip giant is working on "perceptual computing" technology that will sense your emotions and your body language. Here's an inside look.

Shara Tibken Former managing editor
Shara Tibken was a managing editor at CNET News, overseeing a team covering tech policy, EU tech, mobile and the digital divide. She previously covered mobile as a senior reporter at CNET and also wrote for Dow Jones Newswires and The Wall Street Journal. Shara is a native Midwesterner who still prefers "pop" over "soda."
Shara Tibken
4 min read
Anil Nanduri, an Intel executive working on perceptual computing, demos technology that senses users' individual fingers. Shara Tibken/CNET
SANTA CLARA, Calif. -- If the next big wave in devices turns out to be gestures and eye tracking, Intel wants to be ready.

Intel is the king of PCs, but it hasn't always been ahead of evolving innovations. Its processors power more than 80 percent of the world's computers and the vast majority of its servers, but Intel has made little headway in smartphones and tablets. To spur interest in PCs again, as well as persuade more mobile device makers to use its chips, Intel has devoted significant resources and efforts to something it calls "perceptual computing."

Perceptual computing may sound like a jargony, marketing term, but it does just what it says -- it uses the senses to help technology interpret what's going on around it. Those features, such as gestures, facial recognition, and voice recognition, should all make devices more "natural, intuitive, and immersive," says Anil Nanduri, one of the Intel executives in charge of the company's efforts in perceptual computing.

Inside Intel's perceptual computing lab (pictures)

See all photos

The goal is getting "sensory inputs that make [computers] more human like," Nanduri said. "Once you give computers the ability to assess depth, a lot of wonderful things can happen."

Devices will be able to sense emotion and detect a person's biometric data simply using a camera. They'll be able to carry on conversations with users and understand context -- or what "play me some jazz" means -- instead of simply following commands. Computers will be able to pick out individual fingers instead of simply recognizing an entire hand or the fact that a person is present. And they'll create more immersive augmented reality, such as digital versions of children's pop-up books.

In the case of Intel, the company is placing particular emphasis on vision and teaching devices to recognize depth. That's made possible through 3D cameras. The company has partnered with Creative on 3D cameras, which should show up integrated into devices such as PCs and tablets in the second half of 2014.

A big pitfall for companies like Intel is the hyper focus on speeds and feeds, making technology that's the most powerful without necessarily considering all the ways it might be used. For perceptual computing, Intel says it's starting with software and users first and then moving to the hardware.

To do that, Intel released a software development kit last year to get developers interested in the technology. Since that time, the SDK has been downloaded more than 26,000 times. Intel is so serious about perceptual computing that it has even sponsored contests -- with $100 million in prizes -- to get app developers interested in the technology. Intel will announce the latest crop of winners soon.

"For the users, what am I getting for it?" Nanduri said. "That's why we started a year early, focusing on the ecosystem more so than talking about bringing this into hardware or a device."

Yuriy Kozachuk, an application engineer in Intel's perceptual computing lab, demos technology that tracks facial expressions and translates them to characters in a game. Shara Tibken/CNET
But now Intel believes the ecosystem has advanced enough that it's time to talk hardware. Devices will show up next year that contain elements of Intel's perceptual computing efforts. And it hopes all of those will use its chips. Technically, some features could be possible using chips such as those from Qualcomm. However, Intel says the amount of horsepower needed to run the features smoothly will require its higher-end chips.

Initially, the perceptual computing features will only work with Intel's Core line traditionally used in PCs and some tablets, not its lower-power Atom line used in mobile devices. However, the company plans to eventually make the features run on its more energy efficient processors, and it's also adding accelerators, tools, and graphics to its chips to take advantage of the perceptual computing capabilities.

"We're already thinking ahead and looking at the use cases people need two to three years out from now and putting them into our silicon," Nanduri said.

Some elements of perceptual computing have already shown up in products. The Kinect for Microsoft's Xbox is one example, as are Siri and Google Now for voice recognition. However, Intel says it's taking those a step further by focusing on short-range interaction of less than a meter. That means the technology needs a very fine level of recognition, with the ability to pick out specific fingers instead of just noticing an arm or if entire person is present.

But it still will be a challenge for Intel to make features that are truly useful and not just gimmicky. Intel acknowledges that gesturse and other features won't be ideal for all instances. Computer users, for instance, won't be making slideshows by waving their hands in the air. But they might use gestures when showing the slideshow to friends.

Gaming, in particular, is one area where perceptual computing could really take off, Nanduri said, as well as education and related fields. And it's not just about PCs. This technology will show up in a wide range of devices in the coming years, he said.

The company is sure to provide more details and demos in January at the Consumer Electronics Show in Las Vegas.

"Perceptual computing is about everything and is device agnostic in many ways," Nanduri said. "It's going to be everywhere."