ZDNet AnchorDesk Editorial Feature

To print: Click here or Select File and then Print from your browser's menu

	--------------------------------------------------------------
	This story was printed from Anchordesk,
	located at http://review.zdnet.com/AnchorDesk/.
	--------------------------------------------------------------

Personal robots: Why they're nearly here

By Patrick Houston: Editorial Director, AnchorDesk

Friday, January 24, 2003

The Aibo turns four in a few months--or 28, if you measure the age of Sony's mechanical pooch in canine time. I didn't expect it to live so long.

Even though Sony sold out the first 5,000 Aibos it produced within 20 minutes, I've always considered the gadget--with top-end prices pushing $1,500--as a novelty for the Hammacher-Schlemmer set.

But then, as part of our recent search for the Next Big Thing in personal tech, I talked to robotics experts Victor Matsuda, Helen Greiner, Una-May O'Reilly, and Hans Moravec. Matsuda is the Sony VP in charge of the company's U.S.-based Entertainment Robot America division. Greiner is an entrepreneur. O'Reilly and Moravec are researchers from MIT and Carnegie Mellon University, respectively.

After chatting with them all, I believe that, starting this year, more and more of you will be welcoming truly utilitarian robots into your homes, where they will vacuum the floors, watch the premises, serve as "personal agents," and otherwise help you live your life more efficiently.

Yes, robots are no longer just the stuff of Star Wars sequels anymore. They won't be relegated to the factory floor. Robots are getting real.

THE SONY AIBO has grown increasingly sophisticated since its introduction in June 1999. Its top-of-the-line model, the ERS-220A, runs on a 64-bit RISC processor and 32MB of SDRAM, and has an array of image, pressure, distance, and temperature sensors. Late last year, Sony began offering enhanced software that allows the Aibo to dance in time to music it "hears," recognize its owner's voice and face, and find its way to its recharging station when it runs low on juice.

Matsuda hinted (Sony is notoriously circumspect) that 2003 will bring more advances to the Aibo platform, advances that will position the robot for more practical tasks. What would Aibo, the personal agent, do? The possibilities, according to Matsuda, might entail having a wirelessly networked Aibo recognize various members of the household and offer each appropriate reminders.

Little Aibo might walk up to Mom and say, "You have mail." It might scamper to Dad and say, "Don't forget the kids have a dentist appointment Tuesday at 10 a.m." Or it could just watch Junior doing his homework in his room--and transmit the scene via the Web to Mom or Dad at work.

Don't forget who's painting this picture. Sony doesn't dabble in niche markets; it deals in mass markets--and owns the muscle to create them. Cautious by nature, it also tends not to get too far out ahead of consumer demand. If Sony has identified an opportunity in personal robotics, you can bet that it sees a big one coming--and soon.

Pat tests iRobot's robotic floor vacuum and talks about why robots are a perfect fit for the home.

Watch now

Sony's Jon Piazza demonstrates the latest Aibo, including upgrades to its voice and face recognition software--and a new skateboard.

Watch now

SONY ISN'T THE ONLY bellwether on which I'm basing this forecast. Helen Greiner is a former MIT artificial intelligence researcher from Hungary who 13 years ago co-founded iRobot Corp. Her goal: To bring robots to everyday life. The company has since developed robots for industry and the military. In 2000, its engineering expertise helped bring Hasbro's My Real Baby doll--which used AI software to mimic emotions--to life.

But iRobot toiled in relative obscurity until last September, when it stormed the market with a robot that sucks: a $200 smart vacuum cleaner called the Roomba, which, while cleaning a room, keeps track of its location using some pretty sophisticated algorithms. The Roomba has since reaped many awards and captivated a large crowd when Greiner demonstrated it at the Consumer Electronics Show (CES) recently. (See the video of her demo here.)

Buoyed by Roomba's success, iRobot is planning a "whole category of robots that can help people take on household tasks," Greiner says.

Other personal robots are making their way into commercial production, too. Two weeks ago, Sweden's Electrolux debuted its own robotic vacuum, called the Trilobite. Evolution Robotics, a small company in Pasadena, Calif., used the just-concluded CES as the backdrop to debut its ER2, which was designed for tasks such as home security, reading to the kids, and video conferencing. Last fall, an Italian company, Zucchetti Centro Sistemi, introduced Ambrogio, a robotic lawnmower.

WILL ANYBODY use these things? They already are. And not just anybody. Take, for example, Una-May O'Reilly. She's the mother of three kids under the age of three. She owns a Roomba. She loves it.

But what makes her testimonial so notable is that she's not just a working mom with her hands full. She's also a research scientist at MIT's Artificial Intelligence Labs, where she's been part of several robotics projects. She sees the Roomba and the Aibo for their larger significance: They are "heralding the advent of low-cost robots to mass markets."

Hans Moravec sees it that way, too. He's a principal research scientist at the Robotics Institute of Carnegie Mellon in my old hometown of Pittsburgh. He's been experimenting with robots since childhood. And he offers the most cogent affirmation I found as to why robots are poised to enter our lives now, in real ways--and no longer in just imagined or experimental ones.

It's not the software or the sensors or the engineering. Instead, Moravec says, processors have gotten powerful enough to make it "just barely possible" to create useful robots. He figures that chips capable of 100 million instructions per second (MIPS), costing a few hundred dollars, can handle the complexity of perception and movement roughly equivalent to that of an insect.

With 1,000 MIPS, or about what a 1GHz Pentium can deliver, he's been able to create robots capable of assembling a highly detailed 3-D grid map of their surroundings. By updating the map every second, these robots can reliably move about a room, pretty much in real time.

By Moravec's reckoning, we'll have robots capable of mimicking the complexity of a lizard's nervous system by 2005 and a monkey's by 2015.

Moravec also believes that much of this progress could hinge on something other than technology: commercial success. If a product reaps revenue in any significant amount, we could be seeing personal robots sooner. If not, later.

But then again, that's why I wouldn't be surprised to see the Aibo reach 28--and I'm not talking dog years.

What do you think? Would you buy a robot to do the vacuuming? What would your ideal house-robot be able to do? TalkBack to me!