Intel Research seeks ways to put computer intelligence into everyday objects

For James Landay and his fellow researchers, the most interesting computers look nothing like the one on your desk. They operate inside bracelets, gloves and bathmats.

Landay, director of Intel Research in Seattle, leads a team of 17 people quietly applying their collective brainpower toward an idea called "ubiquitous" computing.

Intel, the Santa Clara, Calif.-based semiconductor giant that helped launch the PC revolution, is paying them to take the revolution outside the box. Depending on the project, that can mean the streets of Seattle or the kitchens of local grandparents.

The idea is to put computer intelligence into everyday objects that people interact with naturally, so the objects become "smarter" and can anticipate what people might want them to do.

"My cellphone should know if I'm in a restaurant or a movie theater and vibrate instead of ringing," Landay said.

Ultimately, making more smart objects and scattering them about help Intel sell more chips.

Intel opened a lab in Seattle three years ago to work near the University of Washington and collaborate with UW's Computer Science and Engineering Department. Intel also put labs near the University of California, Berkeley, Carnegie Mellon University in Pittsburgh and the University of Cambridge in England. All four are among the top computer-science schools in the world.

Universities, after all, have provided the fertile ground for technology breakthroughs to get started, including the Yahoo! Internet directory, the Napster file-sharing system and the Google search engine.

"The idea is to see things earlier by being closer to the ground," Landay said. "We're looking for things that five or 10 years out could be real game-changers."

Intel's Seattle lab has three major projects under way, with plans to grow to 20 full-time researchers, along with a dozen interns in the summer. Eventually, their work might end up in a new product or even be spun off into a new company. Work from the UC Berkeley lab made its way into a Silicon Valley start-up called Crossbow, whose investors include Intel Capital.

The new labs are separate from Intel's corporate research and represent only a tiny fraction of its $4.8 billion annual R&D budget. Other tech giants like Microsoft, Hewlett-Packard and IBM also have major research and development labs, but corporate labs have a different agenda.

"Anything that happens there, the corporation is going to own," said Landay, who came to Intel's Seattle lab from a faculty position at UC Berkeley. "Often university researchers aren't going to want to work with them because they have no leverage. Here it's on a level playing field."

The new model is open collaboration and sharing of information, with the fruits of projects jointly owned by the university and Intel.

"The Intel labs are absolutely unique among corporate research labs in that their explicit purpose is to collaborate with the neighboring university," said Ed Lazowska, who was chair of the UW Computer Science and Engineering Department when the lab was established. Lazowska said Intel's presence helped him lure talented professors to UW, since many faculty members split their research time between the university and the lab.

"You get a free exchange of ideas," said Jeffrey Hightower, a recent UW graduate who joined Intel Research in August. "People from the university can easily work with me without a lot of heavyweight intellectual property issues getting in the way."

Hightower said he found the hybrid model so appealing that he turned down job offers at Microsoft Research and the University of California at Irvine.

Project called Place Lab

From the top floor of a new six-story building in the University District, he and other Intel researchers are working on a project called Place Lab.

Place Lab software helps a mobile device find its location by listening for radio beacons in the environment, such as wireless access points that provide Internet service and signals from cellular-phone base stations. Each access point has a unique number.

Place Lab can identify the origin of the radio signal by comparing it with a set of access points stored inside the mobile device, pinpointing the location within 50 to 75 feet.

That's not as precise as current Global Positioning System technology, but Place Lab works indoors and in dense urban landscapes, where GPS doesn't, said Intel researcher Anthony LaMarca, who heads the project.

Future software applications could boost the power of Internet services like maps and geographic searches. Without having to type in an address, someone traveling with a mobile phone or laptop can easily find the nearest bookstore or gas station, for example.

Researchers also have tackled the thorny problem of privacy in their design. Unlike mobile phones, which constantly broadcast their location by communicating with cell towers, Place Lab stores its maps in the device rather than on a network server and listens passively for radio signals without sending any data back out.

So far, Place Lab software has been downloaded about 3,000 times. Students at UC Berkeley and UW have used it to create new instant-messaging applications that let people show their current whereabouts to others on their buddy list.

If computers can figure out where people are, they also should be able to guess what they're doing.

To make his point, Landay shows stills from the movie "Memoirs of an Invisible Man." The main character might be invisible, but whatever he touches gives him away: a coffee cup suspended above its saucer, or a telephone receiver dangling in midair.

Technology still can't make people invisible, but sensor networks are getting better at understanding human behavior.

"If you know what objects people are using, you get a pretty good idea what they're doing," Landay said.

Landay slips on a black leather glove that has a built-in device called a radio frequency identification (RFID) reader. The reader, the size of a key fob, picks up wireless signals sent by tiny sensors.

The nickel-sized sensors, called RFID tags, are affixed to all kinds of household objects, including cups, bowls, a tea kettle and a toothbrush. Whenever an object is picked up, sensors register the action.

Researchers link the action to a computer program with steps needed for 20,000 different activities, such as making tea, washing your hands and brushing your teeth. If the right steps are followed, the program infers that a person has completed the task.

The project is designed to help people with Alzheimer's disease live independently by allowing family members or caregivers to keep track of their daily routines remotely. Intel's Seattle-based researchers contributed their findings to a project at Intel Proactive Health Research lab in Hillsboro, Ore.

The technology also could apply to training programs. The Seattle lab is working with the UW Medical School to see whether sensor networks can help determine whether medical students have used the right tools in the right order as they simulate operating-room procedures.

Landay's most recent project, Digital Simplicity, is aimed at improving health through technology. The concepts include a wearable computer that measures activities and exercise throughout the day and displays the results in a ring.

Kristi Heim: 206-464-2718 or kheim@seattletimes.com

An experimental touchscreen records the routine actions of a hypothetical nursing-home resident observed with the help of radio frequency identification tags attached to household objects, such as cups, bowls and a toothbrush. (KEN LAMBERT / THE SEATTLE TIMES)