When to Stop?

Active safety technologies — which can identify an approaching object and apply brake force on behalf of an inattentive driver — have vast potential to prevent injuries and deaths on the world's roadways.

Their anticipated rollout extends, literally, from bumper to bumper, as a combination of radar, laser and camera-based vision systems survey the area in front of, behind and beside a vehicle as it changes lanes, backs out of the driveway and motors down the highway.

But before the technology arrives in production vehicles, OEMs and suppliers must answer some fundamental questions and enable these devices to answer them as well, with split-second accuracy.

Active safety systems need intelligence to identify which objects are worth stopping for and which are not. For instance, forward detection systems can spot small objects such as a soda can or a tree branch. Most drivers would roll over either one, assuming it's small enough.

But what about a squirrel? An animal lover would step on the brakes, but someone else in a bad mood might not.

It's too early to say whether such a system could reliably detect a squirrel and apply braking pressure in a bid to spare its life, but a number of suppliers and auto makers are engaged in product development efforts to address such issues.

Germany is becoming the center of expertise for active safety, as luxury brands such as Mercedes-Benz, BMW and Audi have been early adopters of adaptive cruise control (ACC), which provides a gateway to more advanced object detection.

In late May and early June, three German companies — Volkswagen Group, Robert Bosch GmbH and Continental Automotive Systems — hosted media events to demonstrate their progress in the sector. Although the events were held separately, the technology and approaches discussed are remarkably similar.

Continental advances with “sensor fusion,” which allows information gathered from two or more separate sensors or cameras to be processed in unison. A radar sensor can identify a soda can because the metal reflects radar waves back to the car.

But a second device such as a CMOS camera working in tandem gives the system more intelligence by providing additional information about the object ahead. Using algorithms and software, the electronic controller governing the system learns to identify objects the same way the human brain does — by repetition.

“This controller has to learn what is a car, a motorcycle, a pedestrian,” says Peter Rieth, vice president-technology and advanced engineering at Continental. “These algorithms learn from drive to drive. They build up an experience as we have in our brains, although these attributes are learned by the computer.”

The Mercedes S-Class is Continental's first example of sensor fusion because it employs three frontal radar sensors for ACC. If the sensors identify an object in front of the car, they warn the driver.

“It could be, the driver is not paying attention. That's a very important feature to wake up the driver,” says Karl-Thomas Neumann, president of Continental Automotive. “You are the driver — it is your responsibility.”

Once the driver steps on the brake, the system provides additional stopping force.

Suppliers and auto makers must tread carefully to ensure the systems do their jobs without annoying the driver. Thousands of objects may be detected during a short drive to the grocery store, but the driver needs to know only about those that are potential hazards.

Neumann says he is confident Continental engineers will improve these systems, but he says it will be a long time before a vehicle can initiate a fully automated emergency braking event, independent of the driver, due to liability issues.

The production rollout of more advanced forms of sensor fusion for accident prevention should come within the next three years, although Continental has no signed contracts yet, Rieth says.

Continental competitor Bosch is working on the same technologies. Last year, Bosch launched a Predictive Collision Warning system on the Audi Q7 that (working with radar-based ACC) provides a short brake pulse to the driver when it detects an object in front of the vehicle.

The system also could alert the driver by tugging the seatbelt or sounding an alarm, but Bosch's research found an automatically applied brake pulse consistently is the best way to get the driver's attention.

Bosch also has Predictive Brake Assist, which prepares the braking system to apply maximum stopping force once the ACC sensor identifies a situation that likely will require an emergency stop.

The next step for Bosch is Predictive Emergency Braking, which uses a video sensor (in addition to ACC) to do a “plausibility check” to see if an object is worth stopping for, says Edwin Liebemann, Bosch vice president-chassis systems control marketing.

The system will then intervene — even if the driver never touches the brake pedal or reacts to warnings that a crash is imminent — and produces maximum deceleration in reducing the seriousness of the crash. Bosch says drivers fail to react at all in about half of all rear-end collisions.

The Predictive Emergency Braking system, scheduled to debut in 2009 on a European luxury car, also can interpret road signs and traffic signals and even warn drivers if they are speeding.

So will this system spare the life of a squirrel?

“It depends,” Liebemann says. “I would assume that from a squirrel, the likelihood that we actually get back a signal from the radar sensor is rather low. If there was a human being or a horse or a large dog, then we will receive enough information.”

The size of the object to be detected is important, as is its composition.

“Radar beams are reflected by metal objects and dense objects,” Liebemann says, noting an aluminum soda can is just as likely as a vehicle to be detected.

“That's why we need the additional eyes — like video sensors — to determine, ‘do we have to brake or don't we?’” he says.

Bosch also is attempting to connect a vehicle's airbag control units with electronic stability control. Bosch is the world's largest producer of ESC, which uses sensors to detect (and generally prevent) a vehicle from skidding out of control — the most frequent cause of fatal accidents in Germany.

If ESC can notify the airbags that a vehicle is skidding sideways (potentially striking an object such as a pole), the vehicle can prepare for an impact and the airbags can deploy a few milliseconds earlier, helping to protect occupants.

Bosch says the Early Pole Crash Detection system will be ready for series production by the end of next year.

Researchers expect these predictive safety systems to reduce the number of fatal accidents by 35%, says Rainer Kallenbach, Bosch executive vice president-sales, automotive electronics.

Safety devices already are paying dividends. In Germany alone, Kallenbach says the number of fatal accidents annually has plummeted from 19,200 in 1970 to about 5,000 currently, despite a growing population of motorists and the number of vehicles and miles driven.

At its recent event at a test track near Wolfsburg, VW demonstrated a prototype Phaeton equipped with a camera and laser sensors designed to offer “left turn assistance” and “traffic light assistance.”

The system detects the driver wants to turn left and surveys the area ahead. If oncoming traffic is identified, the driver sees a special risk indicator on the dashboard.

The arrow of the turn signal indicator turns red, calling further attention to the hazard. In critical cases, an acoustical signal sounds.

At intersections, where VW says 35% of accidents in Germany occur, the vehicle communicates with traffic signals via radio wave. The optimal (and allowable) speed is shown to the driver so he can pass through while the light is green.

If the system detects the driver is going too fast, it activates an acoustic warning signal, allowing time for the driver to brake. In both instances, the driver has ultimate control — and responsibility.

And some day in the future, VW, Continental and Bosch say vehicles will be able to communicate with each other and with a network infrastructure to identify trouble spots on the road ahead, such as a major traffic jam, a disabled vehicle on the shoulder or a fire truck speeding to an emergency nearby.

For example, if a vehicle needs to brake abruptly due to an accident — or is involved in an accident, itself — it can send this information in fractions of a second to cars that are following, helping avoid collisions.

And if one vehicle's wheels are slipping and headlamps and windshield wipers are switched on, other vehicles nearby can be alerted to inclement weather ahead.

Vehicles equipped with such devices also become roving traffic reporters, sharing information back and forth. Think of it as a vehicular Internet.

Continental's Rieth says he expects car-to-car technology on the road by 2013. The biggest challenge is the phase-in period, in which some cars have the technology and many do not. What good is it to have “smart” cars on the same roads as “dumb” cars?

Continental's Neumann concedes the question is relevant. “If just one car has it, it has no value,” he says. “There must be a number of cars, a certain penetration to really get value from this kind of communication.