New Devices Let Cars 'See' Through Fog, Drive Themselves

Don't expect a robotic vehicle to show up at the door with your latest web purchase or your car to drive itself home tonight. But, a team of researchers led by Electrical Engineering Professor Ioannis Kanellakopoulos has developed what he calls an "advanced vehicle control laboratory on wheels" to test these concepts and others envisioned for vehicles of the future.

The platform is designed to test features which allow the speed of a vehicle to be controlled without the need for any mechanical linkage connecting the accelerator and the throttle (Drive-by-Wire) and alter the vehicle's direction via a computer algorithm, eliminating the need for a steering column (Steer-by-Wire). Also being tested is an innovative ranging sensor that calculates the distance between vehicles and acts as a collision warning device.

In addition, the platform is designed to test innovations which have not yet been developed such as autonomous capabilities that will allow vehicles to drive by themselves.

"We wanted to have the capability to do things that the car companies are not even thinking about so we can do every imaginable experiment," Kanellakopoulos said. The vehicle was also designed to accept additional features when funds become available. One such feature is the ability to evaluate static and active suspension.

Some of the features - such as Drive-by-Wire - are already available in production versions of higher-end vehicles. Others will be available soon.

Steer-by-Wire, for example, is still experimental. One of the challenges it poses is providing drivers with appropriate feedback. For assistance in this area, researchers turned to video game designers. The steering wheel on the test vehicle contains an electric motor which provides resistance equal to that experienced when turning the steering wheel of a conventional vehicle. In fact, the effect is so realistic that "if you bump up against the curb, you shouldn't be able to turn the steering wheel and this motor is powerful enough to keep you from doing so," Kanellakopoulos said.

Eliminating the steering column, according to Kanellakopoulos, would increase safety and performance as well as reduce production costs.

"Imagine a car that doesn't have a steering column," said Kanellakopoulos, who considers the steering column "a major design limitation." Eliminating the steering column would remove one of the more common objects which cause injury during automobile accidents, Kanellakopoulos said. "You don't even need a steering wheel anymore for that matter. You could have a joystick."

This could also reduce the cost of production, because "the more things you can do electronically the cheaper they are to produce," he said. "The economics of scale works much better for electronic components."

Eliminating the steering column would also reduce weight and make the vehicle more fuel-efficient.

Kanellakopoulos uses the history of Antilock Brake Systems (ABS) to illustrate how many automotive advances evolve. This technology was first made available on higher-end vehicles at a somewhat greater cost. It then propagated to lower cost vehicles and is now standard on most cars. But ABS also provides the foundation for other innovations such as traction control. Using sensors at each wheel, ABS prevents skidding by releasing braking pressure to a specific wheel whenever that wheel stops turning.

Most traction control systems use the same types of sensors to detect wheel speed. To correct for over-steer and under-steer, "we apply the brake on individual wheels which provides the necessary torque to turn the car the right way," Kanellakopoulos said.

"And it doesn't cost you anything extra," Kanellakopoulos said. "Once you've got the sensors and actuators in place, then adding more features just becomes a matter of adding more processing power and more algorithms. Both of those are very cheap."

Although initial funding for the project came from a grant from the TRW Foundation, researchers are currently collaborating with Ford and Visteon on a Steer-by-Wire project. (Visteon is a spin-off from Ford, just like Delphi was a spin-off from GM.)

One of the more promising avenues researchers have explored involves the use of an Infrared Ranging via Image Subtraction (IRIS) sensor. IRIS capitalizes on the fact that all vehicles contain retroreflective surfaces. On automobiles, these are found in taillights. Such vehicles as bicycles have similar retroreflectors.

IRIS captures and compares two images: one image illuminated with its infrared beam and the other illuminated only by ambient light. The difference between the two images is the retroreflective surface.

"So you can very clearly see the taillights of the other cars or a reflective bumper sticker," Kanellakopoulos said.

IRIS could help in the development of what Kanellakopoulos calls "smart cruise control."

Infrared Ranging via Image Subtraction (IRIS).

"You select not only the speed you want to drive at, but also the distance you want to keep from the car in front of you. If the car in front of you is moving slower than you are, your car slows down to follow it, so you can use your speed control in city traffic - even on the 405 (one of L.A.'s most crowded freeways)."

But IRIS has its limitations. "It won't see a tree. If there is a child in the street who isn't wearing reflective clothing, you won't see him, either. So it's designed for the highway environment where you won't have trees in front of you or pedestrians," Kanellakopoulos said.

Extremely thick fog and very heavy rain can also foil IRIS. However, Kanellakopoulos said, "The experiments that we have done indicate that it will work even when your eyes do not."

For these reasons, IRIS would complement other systems currently in use on some automobiles. Systems on high-end automobiles use radar sensors.

Radar, however, is not without problems. To remove clutter, most radar sensors separate out and discard signals bouncing back from any objects which are not moving. So even though radar can penetrate fog, it cannot see stopped cars because most radar sensors cannot differentiate between a stopped car and the road. Radar can, however, spot children - which the IRIS system is unable to do.

"But if you've got both of these sensors, you can see the reflection of a stopped car very clearly," Kanellakopoulos said. He envisions such a dual-sensor system for high-end cars, or for trucks "where you want very high reliability, and cost is not the primary issue."

Kanellakopoulos also foresees the use of IRIS sensors alone as a means of bringing these new advanced functions to low-end cars where cost is a primary consideration.

"I think with the gradual introduction of these advanced features, people will eventually start saying 'if it can follow another car or if it can brake on its own, why can't it take me where I want to go?'"

By integrating this technology with the Global Positioning System (GPS), which provides accurate location information, Kanellakopoulos said, "it's conceivable that instead of telling the driver where to turn, the system could turn on its own. It knows where to go and it knows how to actuate the car so it can take the action on its own."

With devices such as these responsible for the safety of the vehicle, there are huge liabilities associated with all of this technology.

Kanellakopoulos believes this will be one of the primary obstacles in bringing this technology to the U.S. market.

"It would be great if my car had these fully automated capabilities. So I can go on the freeway and tell it 'I want to get off in 35 miles.' And it gets on a dedicated lane and this lane goes much faster. I don't have to wait in traffic. I'm polluting less because I'm spending less time sitting idle and it's safer because if my attention gets distracted I'm not fully responsible for the system at that point."

In addition, systems such as these can "double or triple the capacity of existing freeways, by allowing vehicles to travel safely at higher speeds, with less distance between them."

One of the researchers' eventual goals, Kanellakopoulos said, is automating cars.

With the advent of telecommuting and online shopping, Kanellakopoulos said, some may question the need for high-tech automotive innovations.

Anticipating these advances, the test platform is also capable of carrying out experiments in wireless communications and one of the concepts Kanellakopoulos said researchers have considered, but not yet pursued, is remote guidance. The car drives itself, but when it comes to an intersection, it stops and waits for directions from the base station.

"You want the car to have autonomous capabilities, but you want to do some guidance remotely," he said.

"Imagine now if people really didn't want to drive except on occasion, for leisure, but they don't want to go to work and they don't want to go to the stores," Kanellakopoulos said. Shopping from home will put increased pressure on the postal service and other package delivery services. "How are they going to deliver millions and millions and millions packages on time?" Kanellakopoulos asked.

Kanellakopoulos' solution: a fleet of automated vehicles, capable of picking up packages and delivering them to the proper addresses.

Kanellakopoulos foresees automation playing its largest role in commercial vehicles. In fact, he is currently working on a project to automate an 18-wheeler.

"We have a track laid out and we're collaborating with a group from U.C. Berkeley. You get in, touch a key on the computer, and the thing takes off, accelerates, goes through the curves, slowing down as necessary and accelerating as necessary and at the other end it stops. And the driver doesn't do anything."

But before any of these innovations show up on the "street," there's a good chance they will be tested on Kanellakopoulos' "advanced vehicle control laboratory on wheels."

-David Brown