Canesta lets computers

A small Silicon Valley start-up has embarked on an almost mythological effort to give computers the gift of sight.

Tales of secret powers bestowed upon men by Greek gods may have a theatrical edge over Canesta Inc.’s modern-day technology, but the company’s plans to let computers see the world in three dimensions is dramatic in its own right. By combining its chip technology and image-processing software, Canesta has made it possible for a PC or handheld device to look out at the world via a small lens and create a three-dimensional picture of the objects around it, according to the company.

The technology could lead to new uses for computers as well as new tools for end users, from virtual keyboards that float in mid-air to powerful security systems for guarding access to a house or a car.

Unlike standard imaging products that measure only light, San Jose, Calif.-based Canesta’s system takes both light and distance readings of the objects in front of its lens. Light is bounced off the objects and Canesta’s chips measure how far away the objects are, creating a 3-D image of the surroundings. The distance information is gathered and processed in real time and can create a detailed view of the outside world that computers can understand and ultimately interact with.

“Traditional cameras operate in a two-dimensional world that makes it difficult for them to tell the difference between a person and a picture of the person,” said Jim Spare, vice-president of marketing at Canesta. “We can provide a much richer set of information that would be able to show the contours of a person’s face, for example.”

Applications of Canesta’s technology could stretch across a variety of industries, including computing, security, gaming and automotive, he said.

For example, current facial recognition systems typically measure the distances between a person’s eyes, nose and other features but don’t measure their facial contours. Canesta’s technology can be used to create a 3-D image of the face, with each twist and turn of an ear accounted for, allowing for a more precise security system.

Video gamers, meanwhile, could put their skills to the test by interacting physically with games in ways other than just hitting keys on a controller. Gaming companies could mount a lens on their consoles that watched a player’s every move, registering a duck, jump or kick.

Although some of the ideas may be years from reality, Canesta says its technology is ready to take the first steps into the market for mobile devices. The company hopes that by early 2003, some manufacturers will place a lens the size of a fingertip in the ends of their cell phones and handheld computers. The lens could project a keyboard in front of a user that would process hand movements, allowing a person to type “virtually” without the need for a physical keyboard, Spare said. The technology could be particularly useful for small, wireless devices, said one analyst.

“With handhelds today, you have to pull out a pen when you want to enter information,” said Bob Levin, senior industry analyst at Mobile Insights Inc., in Mountain View, Calif. “With their sensors, you could type just by holding your fingers above your handheld.”

Easy-to-use input methods will become even more crucial as handheld devices mature and offer more complicated types of information to users, he said. “You will need to manage very extensive menuing systems,” Levin said.

Canesta’s technology is the first of its kind to be offered at a price low enough to draw the interest of device makers, Levin said. The company estimates that its technology will add “tens of dollars” to the cost of making a cell phone or handheld computer, meaning it is likely to appear in higher end devices priced at US$300 or more, Spare said.

Levin of Mobile Insights called the company “the real deal,” and said future applications of the technology may stretch well beyond what’s currently imaginable.

One day, he projected, users may be able to hold up a cell phone in a crowded room and have the device match faces with contact information, giving a user the names of any friends or associates present at the event.