With a dashboard computer that can call a tow truck, find the nearest Chinese restaurant or read your email, tomorrow's cars will go the extra mile.
As Americans spend more and more time in their vehicles, automakers are looking for ways to make life on the road easier and more worry-free. A team of researchers in the automotive solutions group at IBM's Thomas J. Watson Research Center believes the answer lies in "automotive information systems" — speech-enabled Internet computers that keep drivers in tune with their vehicles and in touch with the outside world.
Much of the team's work has found its way into a demonstration prototype called DAISY (short for Driver Assistance and Information Systems, with a nod to the movie Driving Miss Daisy). "The purpose of DAISY," says team member Anthony Levas, "is to show how technology can provide the driver with real-time information about the state of the vehicle and link the vehicle with a variety of remote services."
DAISY is the latest embodiment of the automotive research that IBM has engaged in for several years. In 1997, the company worked with Delphi, Sun Microsystems and Netscape to outfit a vehicle with a prototype information system that demonstrated a variety of applications enabled by speech processing, wireless communication and other technology.
Automakers went on to adopt some of these features for top-end luxury cars. General Motors' OnStar® system, for example, links motorists to advisers who provide an array of services. If an airbag deploys, the system automatically notifies an adviser, who calls emergency services and gives the vehicle's exact location based on data from an onboard Global Positioning System. OnStar advisers also provide "concierge services," such as recommending restaurants or hotels and reserving concert tickets.
But IBM plans to take the technology further still. "Existing services involve contacting real people, so they have been limited to small numbers of luxury car owners," says Paul Chou, manager of the mobile solutions group at Watson. "The next step is to provide the same services to tens of millions of nonluxury-car owners, and that requires a robust infrastructure to make most of the services fully automatic."
Drivers can expect to see more and more automotive applications, ranging from collision avoidance technology to messaging systems that deliver up-to-date email, news reports and traffic alerts. All the applications will have to be carefully coordinated to avoid confusing the driver.
"IBM is helping automakers meet the technical challenges involved in these complex systems," says Chou. The company is developing the speech recognition and voice synthesis technology that will go into other manufacturers' automotive information systems. It is also providing the pervasive computing middleware that will connect vehicles to the Internet. To that end, researchers at Watson have been collaborating with IBM teams from the industrial sector and the software group.
Already, IBM is partnering with Motorola to accelerate the technology that will help automakers bring a variety of wireless and Web-based services to drivers and passengers worldwide. "Our work with Motorola is aimed at creating an opportunity for automakers to offer a variety of services that will help them develop stronger, long-term relationships with their customers," says Jonathan Prial, director of marketing and strategy at IBM's Pervasive Computing Division.
MOTORIST'S AID
Among the services being tested with DAISY is the reporting of mechanical failures. The system first alerts the driver with an audio icon, or "earcon." The earcon is followed by a display icon that indicates the urgency of the alert. Accompanying the icon is a recorded voice message such as "Excuse me, engine cylinder number two is not firing. See your dealer as soon as possible." If danger is imminent, an alarm directs the driver to stop the vehicle. The system also provides informational messages — such as an explanation of a warning light — that require no action.
When DAISY announces a problem, it also sends information on the car and the problem to a manufacturer's service center. In addition, the driver has the option of placing a call to the center using simple voice commands. After analyzing the nature of the problem, the adviser can offer advice or arrange for service. If a service adviser is not available, the driver will be notified automatically.
At any time, the driver can ask for a report on the vehicle's condition. DAISY responds by gathering data from the vehicle's diagnostic sensors and microprocessors, sending it to a remote server for analysis and displaying a detailed report. The system provides a spoken summary as well. Such a checkup might be useful before a long trip, or if it has been a while since the last service.
Beyond the sort of maintenance and emergency features being tested in DAISY, IBM is developing technology to enable manufacturers to offer email and automated concierge services. For example, packages under development will translate Web content — say, from an online restaurant-locating service or a weather reporting center — into a text format that can be read by a speech synthesizer.
ROAD PSYCHOLOGY
In the course of developing automotive information systems, researchers are trying to ensure that the new technology is compatible with the way people think and perceive. In the past, automakers have tried offering voice systems that provide maintenance reports or remind drivers of open doors or low gas or oil, but consumers often found the messages annoying. The IBM researchers believe voice systems could become less intrusive. "For example," says Barbara Churchill, a business development manager working in the Automotive Solutions group, "they could be designed so that a message occurs only when the driver requests it, by pushing a button."
The design of DAISY spurred additional research on the comprehensibility of speech. Jennifer Lai, a user interaction designer at Watson, and David Wood, manager of the virtual assistant project at Watson, compared comprehension levels of synthesized and natural speech. They found that the synthesized speech was harder to understand and appeared to interfere with short-term memory. "We suspect that synthesized speech may be even more difficult to comprehend when a person is focused on driving," says Lai. She is currently analyzing results from a joint study conducted with the University of Michigan, in which people in a driving simulator listened to different types of messages with synthesized and natural speech.
Then there is the problem of information overload. "We may have information coming in from a collision avoidance unit, from a navigational aid, from a cell phone and from an email program, among other things," Chou explains. The system needs to prioritize and filter the incoming information so the driver can concentrate on the most important task — driving. "To do this," says Chou, "the system must convey the right information at the right time, so that a driver does not, say, get an email message when he's trying to avoid an oncoming vehicle."
Automotive systems will be all the more useful when they can connect seamlessly to other devices, Chou points out. "We expect drivers and passengers to want to take their cell phones and palmtops with them, so we want to be able to integrate these devices transparently with the vehicle's own systems. For example, you could have the car get an address from the address book in your palmtop." In working toward that goal, IBM is contributing to technology standards such as WAP (wireless application protocol), Bluetooth (used for short-distance radio-frequency communication) and AMIC (automotive multimedia interface collaboration).
Eric J. Lerner is a freelance writer who lives in Lawrenceville, New Jersey.
