|BUSINESSWEEK ONLINE : JULY 17, 2000 ISSUE|
|TECHNOLOGY & YOU
Can Oxygen Turn Sci-Fi into Reality?
A massive public-private joint venture is out to make computing as integral--and natural--as breathing
Oxygen is necessary for life, yet normally, we aren't aware of its presence. If an ambitious research project pays off, computing will become like oxygen: ubiquitous, essential, and mostly invisible.
The goal of the five-year, $50 million Oxygen Partnership Alliance is nothing less than the reinvention of personal computing. Today's special-purpose handheld devices and general-purpose desktops will disappear. The handheld component of Oxygen, called Handy21, will be a device that can supply the correct data to whomever happens to be using it and will morph its functions to suit the needs of the moment, acting as a phone, a remote control, or a camera. Instead of desktops, we'll have an environmental computer, called Enviro21, that disappears into the walls of our rooms and can see us, hear us, and respond to our commands. Network21, an intelligent network, will tie it all together (table, page 24).
Instead of people needing to learn how to use computers, computers will know how to work with people. At the same time, the systems will be designed to protect user privacy and security.
For all the advances of computing, this sounds like science fiction. But both the brainpower and the money behind Oxygen are impressive. Scientists from two Massachusetts Institute of Technology centers--the Lab for Computer Science (LCS) and Artificial Intelligence Lab--will work with researchers at a half-dozen companies: Taiwan's Acer (ACRRF) and Delta Electronics, the U.S.'s Hewlett-Packard (HWP), Japan's NTT (NTT), Finland's Nokia (NOK), and the Netherlands' Philips Electronics (PHG). The companies will contribute a total of $30 million over the next five years, with most of the rest of the $50 million budget coming from the Pentagon's Defense Advanced Research Projects Agency.
At the partnership's kickoff meeting on MIT's Cambridge (Mass.) campus in June, participants showed that they understand the limitations of the technologies they must use. For example, computers are a long way from being able to understand random human speech. Unless a computer knows you're talking about travel plans, it is unlikely to make sense of a statement such as: ''I need a flight to Houston next Tuesday.'' (For a state-of-the-art demo, try MIT's voice-based weather service at 877-573-8255.)
Similarly, computer vision can identify faces and follow crude motion, but can't differentiate between a table lamp and a basketball. The solution, says Michael Dertouzos, director of the Lab for Computer Science and head of the Oxygen project, is to give the system ''a thin veneer of intelligence.'' The computers would be just good enough at human interaction to figure out what we want--or ask for more information.
FUNDAMENTAL SHIFT. The driving force behind Oxygen is the realization that hardware and software design are based on outmoded assumptions. The premise of computer engineering, from the 1940s to today, is that computation and storage are scarce resources and that people must adapt their ways of working to the needs of the machines.
Dertouzos argues that it is time to address the real limit on the usefulness of computers: the patience of the human beings who use them. The premise of Oxygen is that computing power and storage today are very close to what economists call ''free goods,'' commodities so abundant that they need not be rationed. The challenge is making better use of this power, which today is used to squeeze a little faster action out of a game such as Quake or, more likely, sits idle in your office.
The most critical element of Oxygen is speech. Speech recognition on the Handy is a matter of good software and lots of computing power. But enabling people to communicate with the Enviro poses basic questions for acoustic engineering. Human beings can easily understand conversations in a room, even if other people are talking and the television is on. Computers lack our brains' ability to extract the sounds we want to hear from a noisy background. And current voice-recognition software is hopelessly thrown off by echoes--the bottom-of-the-well effect common in speakerphones--and by things as simple as a dishwasher running.
A computer can hardly fade into the background, however, if communicating with it requires headsets or lapel microphones. Instead, Oxygen plans to use tuned arrays of mikes that can work with the vision system to focus the computer's listening on a specific individual, but practical devices don't exist yet.
Some of the problems are more mundane. Battery life is an area that has shown very little improvement in recent years, and no breakthroughs are on the horizon. That's a problem for the Handy21, which needs more power than conventional technology can supply. To cope, the Oxygen team has come up with a new processor designed to save power. It will use only as much power as a particular task requires, then go into hibernation when not in use. The chip, called Raw, is in its final design stages, and IBM Microelectronics should be turning out prototypes by yearend.
The decision to base a critical component of Oxygen on an untested processor of radical design illustrates a philosophical underpinning of the project--what Dertouzos calls ''a safe bottom and a high-risk top.'' The early prototypes of the Handy21 will use an off-the-shelf 200-megahertz Intel StrongARM processor. This ''safe bottom'' should give 15 to 20 hours of battery life in a device that includes a touch-sensitive LCD display, a camera, a microphone, and storage. Raw would be the high-risk payoff--smaller, faster, and with longer battery life.
ORWELLIAN OVERTONES. Some of the technical solutions may spawn their own problems, though. The thought of filling my home with cameras and microphones connected to a computer that is in turn connected to the Internet certainly gives me pause. ''This is potentially the world's greatest surveillance engine,'' says Ron Rivest, co-inventor of the widely used RSA encryption system and head of the information-security group at LCS. ''It is important that we are talking about this on Day One. Too often, security is something pasted on at the end.''
While many participants in the Oxygen kickoff meeting found Rivest's participation a welcome sign that security and privacy concerns were being taken seriously, there are more questions than answers about how individuals will be protected. One of the trial applications of Oxygen technology is a system called Guardian Angel, which allows health-care professionals to monitor patients in their homes: what they eat, how active they are, when they take their medications. You might want to make this information available to your doctor, but you'll want strong assurance that no one else can get to it.
On a simpler level, what happens when someone picks up your Handy21, which is linked to your data on the Enviro21 and the Network21? The Handy is supposed to change its personality and switch to the new user's data. To do so, it could use a password--simple, but neither unobtrusive nor very secure. Or it might recognize your fingerprint or your face--technologies that still need a lot of work. No matter what technology is used, there will have to be strong protections to keep private data private and to prevent the system from being used to spy on people. ''We don't have the answer yet,'' admits Rodney Brooks, director of MIT's AI Lab.
In the end, I think the social obstacles may be Oxygen's biggest challenge. Computers that are aware of our existence and that, on some level, understand us, have great potential to make life easier. But they also create a creepy, Orwellian unease. To succeed, Oxygen must not only perfect technologies that barely exist today but ensure our comfort with hear-all, see-all rooms that seem designed to make us uncomfortable. The Oxygen team is aware of this obstacle, but persuading us to welcome such systems into our homes could be more daunting than building power-thrifty chips or programs that understand speech from across a room.
BY STEPHEN H. WILDSTROM
E-mail firstname.lastname@example.org or fax (202) 383-2125
Want to express your own views on a tech&you column or other technology topic? Join the Technology & You online forum.
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
BACK TO TOP
Can Oxygen Turn Sci-Fi into Reality?
TABLE: The Three Components of Oxygen
E-Mail to Business Week Online