Eric Rombokas points a cell phone camera at an object.

Inside the Intel Labs

August / September 2008 By: Dean Takahashi Volume 6 Number 4

Print this Article

E-mail this Article

A $6 billion research and development budget explains why Intel can come up with one of the world's best science fairs. The company recently showcased 70 R&D projects at its annual Research@Intel event. Of those, only three were directly tied to Intel's main business of making silicon chips. The rest were a diverse collection of technologies that could conceivably improve the future demand for Intel's chips.

Justin Rattner, chief technology officer and senior fellow, said the goal of most of the research is to reshape how people interact with computers in the next five to seven years. In a speech before journalists, Rattner said that Intel's research has a long tail, meaning that it starts early and has a downstream impact many years later. He noted that Intel's atom processor is a low-power processor that began with Intel research back in 1999. The idea was to strip down the processor to its bare bones design to improve power efficiency.

A second major payoff from research was the ability to manage a bunch of computers. The vPro technology—€”where corporate IT administrators can use management tools to better manage a fleet of computers—€”came together from different "islands of research." Starting in 2003, the company kept iterating on the technology year after year, making progress that let it launch a full-fledged business brand in 2006.

The third major research effort that is now paying off is WiMax, a long-range wireless networking technology that is generating multi-billion-dollar investments by wireless operators who want to provide broadband internet service in an effort to challenge the hegemony of the telephone and cable companies. While Sprint and Clearwire will be the companies that operate the WiMax networks based on the Intel research, Rattner says that Intel will benefit in the long run if new categories of devices, such as mobile internet devices, take off thanks to ubiquitous broadband.

"We're not an ivory tower organization," Rattner said. "We want to create things that become a part of products in the market. We believe our investments will pay off handsomely in the years to come." Intel has doubled its R&D spending over the last ten years. That has enabled the company to expand R&D in many areas: scientific discovery, digital health, improving the environment, advanced visual computing, digital enterprise, technology for developing nations, mobility and pure technological exploration.

The projects included the "real-time mobile visual object recognition." Eric Rombokas, the researcher on this project, wants to be able to point a cell phone camera, laptop webcam or mobile internet device and have it recognize the object. In his demo, if you point a camera at an object, the display searches for the object in visual memory and then identifies it as something as descriptive as "day old pizza." Rombokas says that if the recognition program can be programmed to identify thousands of objects, then it could cover most of the things people run across in their daily lives.

A related demo was called "location-based services and new input methods," which used GPS and other location technologies to determine exactly in what direction someone is pointing a camera-display. It made use of a "magnetrometer" attached to the display, which could determine if it was pointing up or in a certain direction. As with the object recognition program, this device could bring up a web site with more information describing a monument or landmark building.

Domestic robots that help people with repetitive tasks are another ripe category where improvements are making the once-fantastic ideas more realistic.
Intel researchers from the company's Pittsburgh lab created a "robot barkeep." They made a mechanical robot arm that could grab cups and fill them up with beer or other drinks at a rate faster than a bartender could do. The technology integrates perception, navigation, planning and other technologies, allowing the researchers to get the arm to do a predefined task, such as stacking cups in a dishwasher.

Billions of people still live in poverty and have never even made a phone call. But Intel is trying to study how to get technology into the hands of children in the developing world.

Tony Salvador, who runs the ethnographic research for the Emerging Markets Platform Group, studies how people use technology in developing nations. He notes how car generators are a big source of electricity for many people who live in remote villages. Hence, Intel needed to create a computer that could use such power. The result is the ClassmatePC, a $250 computer that is moving on to its third iteration. Using either Linux or Microsoft software, the laptops have rugged, spill-proof keyboards. Salvador actually awoke his model from sleep mode by throwing it on the ground. The designs include a little carrying case and can link to other ClassmatePCs, including a master network controlled by a teacher's computer. Salvador is excited about future designs, including making use of touch-screen technology just as the new One Laptop Per Child project has done.

Intel and Neusoft showed how they could build a heads-up display, or transparent messages that could be overlaid on the windshield glass without blocking a driver's view of the road. The computer could enhance the driver's ability to react to emergencies by drawing bright-colored boxes around objects in the field of view that could represent risks. Loaded up with cameras and multi-core processors, the car could alert the driver to dangers such as a car weaving in traffic, a slowdown in the cars ahead, or pedestrians in a dangerous collision course. The smart car might even be taught to take preventive measures on its own, improving driver reaction time.

Given the state of the healthcare system, Intel believes that putting new technology into the healthcare system could help it become more efficient and less costly. For instance, companies may be able to cut down on hospital visits or patient monitoring if they can use remote diagnostics to keep a watch on elderly patients from afar. Sensors could monitor whether Alzheimer's patients have taken their daily medicines.

Tom Stroebel, an Intel researcher, showed off a Motient handheld computer with a big display that could be used by paramedics or nurses who serve rural populations. Doctors could view the live data streams from a webcam and recommend treatments to stabilize emergency victims.

Biological testing chips have been around for a while. But Intel is testing a chemical analysis chip that has more universal testing abilities. Most test chips from companies such as Affymetrix can test for a single malady. But Intel's chip uses silicon electrical sensors with something called a field effect device to detect DNA or other chemical traits more easily. Udi Virobnik, the Intel researcher, said that such chips could do a bunch of tests at once. Such machines could eliminate lab testing, letting doctors get the results from tests almost immediately.

Graphics on mobile devices usually don't look anywhere as good as they do on larger screens. That's why some Intel reseachers want to get phones to broadcast over a wireless network so they can project 3-D graphics images on a big screen.
Mathys Walma and Khanh Nguyen said that they could transmit instructions for graphics rather than graphics themselves over a wireless connection to a display. The display would do all of the hard computation necessary to get the image on a big screen.

Dean Takahashi, the lead writer for digital media at VentureBeat.com, reports from Silicon Valley for Innovation.