When the walls have ears. (Business of Technology).

As envisioned by its leading developers, Philips Electronics and Massachusetts Institute of Technology (MIT), ambient intelligence is a marriage between the technologies of ubiquitous computing and social user interface. On the front end are a variety of tiny devices that can hear, see, or feel your presence. Wireless-based networked systems make sense of this data, figuring out who you are and what you want. Then they make things happen, presumably as you would want them.

In a room equipped with ambient intelligence, there might be no need for switches to flick or buttons to push. You just need to say the word--or maybe just look in a certain direction, if the eyeball-tracking camera can see you--and the networked devices will respond.

Earlier this year, Philips opened a permanent facility in Eindhoven, The Netherlands, called HomeLab, to test prototypes of ambient-intelligence technology. Researchers can stay full-time at the facility for up to two months, living with technology such as home entertainment systems that respond to human voice commands. At MIT, a number of projects have explored the possibilities of environments created for machine-human interaction. In the '90s, MIT'S Media Lab built "smart rooms," described as invisible butlers with cameras, microphones and other sensors enabling them to interpret what people were doing and to help them. While Microsoft and other organizations have been experimenting with ambient technologies for years, the aim now is to commercialize these technologies.

MIT's Artificial Intelligence Laboratory has picked up the theme more recently with its Intelligent Room project, which its website describes as "working towards creating environments analogous to those so familiar to Star Trek viewers, i.e., rooms that listen to you and watch what you do; rooms you can speak with, gesture to, and interact with in other complex ways."

What all this work has in common is a desire to push computing into the background, where it would operate behind familiar objects as an unnoticed but effective presence. Computing would be everywhere--in sensors and devices--yet humans wouldn't need to do the things they do today in dealing with computers, like push keys or run a cursor over a screen.

Here's one example of ambient intelligence in action, envisioned by Philips: A tune keeps running through your head. You can't remember the title, but your smart home entertainment system can, with the help of a Philips database tool called Easy Access that recognizes the melodies and voice patterns. All you need to do is hum a bit of the tune, and the system pulls it up for you from a high-density DVD that can hold thousands of songs.

TV offers some obvious ways to apply this type of technology. Instead of punching buttons on a remote, you might just say, "Switch to channel 27," or end your VCR programming woes with commands like, "Record 'The Sopranos' tomorrow night."

The scope of such personal interaction could range much more widely, depending on where the sensors are embedded. Sensors could be installed in floors, for instance, to detect humans by their weight. The walls literally could have ears or heat sensors to pick up your presence and turn the room lights on for you. The walls could also serve as video displays, conveying information as well as absorbing it.

Obstacles to Overcome

But don't throw away your remote just yet. Some serious hurdles, both technical and economic, have to be surmounted before your typical household appliance becomes like one of the family. Ambient intelligence relies on technologies that are far from fully developed, with no agreed-upon standard yet for operating systems. The size of the potential market for applications, especially in the home, is far from clear. In fact, say some experts, the home may be the last place ambient intelligence takes hold. They see greater near-term commercial potential for it in automobiles and public spaces such as highways, airports or office parks.

"The main technical hurdles are making computers see, hear, speak and understand language," says Robert Laddaga, a research scientist at the MIT Artificial Intelligence Lab. "There are also significant hurdles in power--batteries, low-power devices and providing power lines wherever sensors need to be located." Dick Muntz, a University of California, Los Angeles. professor of computer science who has studied intelligent sensor-based systems in transportation and schools, says the noise and uncertainty in the information from sensors creates a major data-processing challenge. "There is the basic technology problem of how you interpret the environment--such as recognizing faces and human gestures," Muntz says. He also cites the problem of powering sensors and systems, and sees this as one reason why such technology may develop more quickly in automobiles, which have battery power for any installed devices.

On the other side of the ledger, there has been plenty of progress in recent years in technologies such as sensors and wireless, which are needed to make "intelligent" spaces possible. Ed Moran, director of Products and Innovations for Deloitte & Touche's TMT Group in the tri-state area, notes that sensors can be used in more ways as they become smaller and cheaper. And as wireless coverage becomes universal, he says it will be easier to keep track of people as they move within and between buildings, and on the road.

Moran suggests a use for sensors at home that might catch on: Embedding sensors in the floors and walls so that parents can keep track of kids and keep them away from places like the pool or the liquor cabinet (in this relatively low-tech version of ambient intelligence, the children would carry electronic cards to make tracking easy). "People would pay for that. Insurance companies would pay for that;' says Moran.

On the whole, Moran sees sensor-based wireless networking as more useful and marketable in public realms, especially in studying traffic. This is not just to ease congestion, but also to track the movements of individual cars (and drivers) for commercial purposes, like figuring out where to locate drive-through restaurants. The infrastructure is already in place to analyze this data, he says. More sensors--and cheaper sensors--are needed now. "With sensor technology getting cheaper, we can start collecting real-time data on people's location. Then the reverse is true-- we can make things happen remotely, like adjusting traffic lights to change the traffic flow."

At Microsoft, where researchers have experimented with ambient-intelligent devices for the company's EasyLiving project, the focus has shifted to larger-scale intelligent spaces, including office buildings and public places such as corporate campuses and airports. Steve Shafer, a senior researcher in ubiquitous computing at Microsoft Research, says he is working on the technology of "smart buildings," which are "able to deliver location-based services depending on where you are indoors." An example might be a sensor-based system in an airport that determines your exact location by your wireless device and can tell you how to get to baggage claim or the nearest restroom.

Convenience is Key

What would. life be like in a fully ambient-intelligent environment? MIT'S Laddaga says the answer "depends greatly on how good we are at solving the problem of good human interfaces." Take a faulty furnace, for example. "When your furnace shows the 'blue screen of death' and, upon rebooting, you find you need to reinstall the home OS and then reinstall each of the updated drivers (in the proper order, but the proper order is a secret), then life would be hell," he says. "On the other hand, if the home OS could diagnose your furnace problem and tell you that it had already informed the furnace service people, in accord with your programmed wishes, that would be great."

In other words, living with ambient intelligence will be an inviting idea only if those "intelligent" systems arc really as smart--and reliable--as they promise to be. And, if they don't create new demands on the user's own brain. "Technologies that gain ground fastest do two things: They make things easier for people and they save money," says Deloitte & Touche's Moran. Ambient intelligence has to go beyond the phase of amazing technical feats, like CD players that can name that tune, to systems that justify their cost by making life more convenient.

But that, of course, is a challenge faced by computing in general. The work being done now on ambient intelligence should push computer technology further toward becoming truly friendly to the people who use it. As Laddaga notes, "The real payoff from ambient intelligence will be making computers that can live in our world--perceptually enabled computers that can hear and see what we hear and see, that can understand when spoken to, and that can explain their actions and negotiate their future behavior."

Bill Ribaudo is the New England managing partner for Deloitte & Touche's Technology, Media & Telecommunications (TMT) Group (Boston, Mass.)

www.deloitte.com