You’re cruising along at 65 miles per hour, faithfully adhering to Interstate 95’s speed limit. Suddenly the limit drops to 55 miles per hour, as it does in some locations, but you don’t notice. Thwack! You get a ticket, even though you thought you were obeying the law.
If only someone you were riding with had seen that 55 m.p.h. sign. If only your car had somehow “seen” it, and been technologically advanced enough to have slowed down on its own.
Science fiction? I think you know better. Onboard car cameras are all the rage, snapping pictures and video of what’s in front or behind your car, or in your blind spots. A computer instantly reviews the images, spots potential pitfalls, and offers help when you most need it.
If the camera sees that you are veering out of your marked lane, your car can warn you or take more evasive action, like automatically redirecting your steering wheel to a center point.
Back-up cameras scan the area behind your trunk for objects that should not be there, such as a young child. If you don’t hit the brakes, your car will for you.
‘It can then figure out where in the lane markings the vehicle is.’
Cameras can detect how fast the car in front of you is moving, and adjust your cruise control accordingly. They can spot stop signs, and alert you before you run through them. And they can recognize posted speed limits, automatically decreasing your speed as necessary.
I paid particular attention to on-board cameras while perusing dealers’ exhibits at the New England International Auto Show this winter, spotting tiny lenses in windshields, trunks, and front grilles and underneath side mirrors. But I left the show wanting to know more about them — not only what they did, but how they work.
In early April, the National Highway Traffic Safety Administration announced an initiative to get car makers to make car cameras, and related safety technology, standard items on future vehicles. So I sought out Andy Whydell, car-camera guru for TRW Automotive, one of the world’s biggest automotive suppliers and a leading car-camera maker.
From the number of megapixels in a typical car camera, to the machinations of “lane-departure systems,” Whydell explained it all, offering his assessment on what’s currently available in the United States, and what’s coming soon.
Whydell, who is senior manager of product planning for TRW’s global electronics division, said there are primarily two types of car cameras.
A “passive” camera serves as an extra set of eyes for a driver, relaying video to the dashboard or rear-view mirror. Your typical back-up camera is a passive camera, Whydell said.
Higher-end vehicles are now starting to feature “active” cameras as well. An active camera feeds a rapid succession of images — between 15 and 30 images per second — to a microprocessor, which tries to interpret them.
“At the very simplest level it can take a picture and figure out, ‘Is it light or dark out? Is it daytime or nighttime?’ ” Whydell said. “If it detects that it’s dark and the brightness is below a certain level, it can turn on a vehicle’s headlamps for you automatically.”
But your car’s primary camera — typically mounted in the base of your rear-view mirror, but facing forward — can look for much, much more. It locates the horizon by contrasting the brightness of the sky with the darkness of the ground. It finds the traffic lane you’re in by scanning the bottom of each image for white markings 9 to 12 feet apart.
“It can then figure out where in the lane markings the vehicle is,” Whydell said. “Am I centered in the lane markings, or off to one side?”
Based on that feedback, the car’s “lane-departure” or “lane-keeping” system can issue warnings or actively correct your course.
If the camera’s microprocessor is really robust, it can scan images for shapes or patterns other than lane markings, such as stop signs and speed limit signs.
“Speed limit signs are typically posted in a standard location, on a typical height on a pole, and in a certain position. So what the camera can do is look in certain regions of the image” where they’re expected to be, Whydell said. “It’s essentially a pattern-matching exercise. It has a library in its memory of what the standard road signs look like. If it sees a sign that matches the reference image it has in its memory, it kind of ties the two together and says, ‘Bingo! I see one of those kinds of signs here.’ ”
Cars manufactured in Germany and France are already using such technology to recognize speed limit signs, Whydell said. In some European vehicles, sign recognition is tied into automatic braking: if the camera sees a stop sign or a speed limit and knows the car is going faster than it should, it will automatically slow the car.
Your car’s forward-facing camera sees everything you do — sometimes more, Whydell said. Just like you, it starts looking for objects in front of the car, furiously trying to match whatever it sees to a similar image embedded in its memory.
A typical camera might have 5,000 images stored in its memory chip of everything from trucks to motorcycles to pedestrians to deer.
Once the camera has locked in on a target — “yes, I see a car” — its microprocessor determines just how far away it is from your vehicle by analyzing its location relative to the horizon and other things it sees.
Again, such calculations aren’t static. They’re being performed up to 1,800 times per minute. If the car in front of you suddenly slows down,and you don’t, the computer’s going to foresee a collision.
“Based on a certain threshold, like the lane-departure warning, the camera could provide an alertto the driver — either a buzzer or beeper or flashing lights or both — to say, ‘Hey wake up! We’re about to run into the back of somebody,’ ” Whydell said. “We call that feature forward-collision warning.”
At the car show, I saw a few luxury cars boasting “adaptive cruise control.” Such systems automatically slow down vehicles, when set to cruise control, that get too close to the car they are following. When the road’s free, they automatically speed up the vehicle to the anointed cruising speed. Adaptive cruise control is often based on the camera feedback, Whydell said.
Your car might also analyze camera images to trigger “in-city” braking, or “automatic emergency braking,” which is when your car comes to a complete stop on its own before hitting an obstacle — a shopping cart, a parked car, or a person — while at a very low speed. When a rear-view camera is tied into the system, it’s a rear-end collision avoidance system.
Could cameras stop cars traveling much faster — even at highway speeds — when drivers fail to react? The surprising answer, next time.