Similarly, the Volvo S60's new pedestrian detection system combines radar (to gauge the distance and trajectory of moving objects) with a camera (to determine whether they're human). If a pedestrian suddenly steps out in front of the car, the system applies the brakes automatically.
The system includes approximately 10,000 stored images of what pedestrians "look like," according to James Hope, a technical media representative for Volvo Cars of North America in Irvine, Calif.
"The camera compares the stored images to what it's seeing in front, along with looking for human-type movement -- the arms swinging, the shoulders and head, the legs moving -- to determine if it's really seeing a person or something else. The system wouldn't recognize an animal or a baby stroller, but it would recognize the person pushing the stroller," Hope says. Watch the video below to see it in action.
See Volvo's pedestrian detection system in action. Click arrow button to play video. (Adobe Flash is required. Some browsers may require two clicks to start the video.)
These collision avoidance systems require a wealth of data about multiple driving conditions, which allows the computer to factor in every possible variable when deciding upon the appropriate response. As with crash tests, computer simulations have proved key in gathering this data. Volvo tested the S60's pedestrian detection system over 500,000 km of simulated driving in a virtual world, according to Thomas Broberg, a senior technical adviser for safety at the Volvo Car Safety Center in Gothenburg, Sweden.
To generate the data required for better simulations, Volvo has launched 100 test cars in Europe that are collecting driving-condition data over three years, equipped with video cameras and eye-tracking sensors that will aid further research and simulation. The project will collect about 3 million km of live driving data and will help uncover the results of impacts that are less common than front and rear collisions. Broberg says this data is required if Volvo is to reach its zero-fatality goal.
"There will be a combination of protective enhancements and collision avoidance," says Broberg. "This will be a groundbreaker for safety, as soon as you get communication between vehicles and between the vehicle and the infrastructure within the next 10 years, if not sooner."
The ultimate goal? Broberg says cars need to get better at performing actions that rely on human interaction today -- anticipating what the driver wants to do (say, by evaluating head movements), reading road signs and adjusting speed based on driving conditions. "The car will know its state, that the road is slippery and [the driver] seems tired -- there is something coming up in the road, so the car will act evasively," says Broberg.
Of course, a looming challenge for cars that rely on computers for their safety is that computers are not 100% reliable. Car companies address this problem by creating programs that check and recheck the vehicle state many times in a short period. For example, a Mercedes with blind-spot detection sends out multiple signals to verify that it's safe to make a lane change; the car also checks the state of the blind-spot detection system itself and warns the driver if it's not working properly, according to a company representative.
Although we're unlikely to ever see a car that results in absolutely no fatalities from car crashes, it will be an enormous achievement for the auto industry if it even gets close to that goal. According to the Department of Transportation, 37,000 people are killed annually in crashes on U.S. roadways; if even a fraction of those fatalities could be avoided, we'd be putting technology to its best possible use.
If you missed it: See an interactive graphic showing advanced safety features in current and future cars.
Related Stories
- Car tech: Taking drivers' helpers for a spin
- GM unveils 2-wheel, self-driving concept car
- Ford takes Google Maps for a ride
- MIT researchers developing robotic driving companion
- Researcher: Self-driving cars could save U.S. auto industry
John Brandon is a veteran of the computing industry, having worked as an IT manager for 10 years and a tech journalist for another 10. He has written more than 2,500 feature articles and is a regular contributor to Computerworld.