The car that can read road signs
By Simon Hooper
The road sign recognition system works by detecting symmetrical shapes.
What has been the most significant invention or discovery of the past 50 years? VOTE NOW
Find out how technology, science and invention are shaping your world with Explorers
DESIGNS FOR LIFE
James Dyson on the Mini as a design classic
and other leading figures on the inventions and discoveries that changed their lives.
(CNN) -- Inattentive drivers who fail to pay attention to road signs could soon find their cars picking up on their bad habits.
A new system designed by an Australian team at the National Information and Communications Technology Australia (NICTA) laboratory in Canberra enables cars to read road signs and alert drivers who ignore them.
The driver assistant system (DAS) was unveiled at the International Conference on Intelligent Robotic Systems in Sendai, Japan, last week after performing impressively in preliminary tests.
The technology uses a small camera mounted on the rear-view mirror that monitors the road ahead. A dashboard computer then checks images for symmetrical shapes and compares them to a list of road signs.
While previous attempts at automated sign recognition were based on the detection of colored patterns, the NICTA team found shape recognition to be a more reliable system for picking signs out of cluttered scenes and in varying light and weather conditions.
When the computer detects a speed limit sign, another connection to the speedometer tells it if the driver has slowed accordingly. Two cameras mounted on the dashboard also monitor the driver's gaze, triggering an alert if a sign appears to have been ignored.
The technology has potentially significant implications for road safety, with 30 percent of road fatalities estimated to be caused by driver inattention.
"NICTA's aim is to produce driver assistance research that can make driving a safer experience by supporting the driver, such as by keeping track of what the current speed limit is or warning about a stop sign that they haven't reacted to," NICTA researcher Nick Barnes told CNN.
Backseat Big Brother
While some drivers might consider DAS to be a backseat Big Brother, NICTA prefers to compare the device to other driving aids such as anti-skid braking or cruise control. Their research also highlights the common example of inattentive drivers alerted by passengers to cars emerging from side roads or passengers crossing the street.
"The challenge is to make driver assistance relevant to the driver and support them, rather than taking control away from them, or flooding them with information," says Barnes.
"Some drivers may be inclined to switch any such system off, but if it saves them from a big fine by pointing out a speed sign they had missed they may be more appreciative," adds Luke Fletcher, who developed the driver gaze detection technology as part of research towards a PhD.
"Even good drivers find themselves on unfamiliar roads wondering what the speed limit is."
But does DAS herald the dawn of a new age of fully-automated motoring, in which families can settle down to watch DVDs as their vehicle carries them safely to their destination?
Barnes says technology could one day make such a scenario a reality, but adds that there will always be some human element involved in driving, such as in choosing between a scenic route for a gentle Sunday drive or the quickest route for a busy morning commute.
Humans are also more capable in reacting to unexpected events on the road than rigidly programmed automated systems.
Fully automated cars
"There will be a time when, technologically, fully automated cars will be possible," says Barnes.
"But NICTA's aim is to produce systems that are relevant today, and currently full automation in the full range of road situations is not possible. We see the best approach to this as being through supporting the driver."
The prototype NICTA vehicle performed well in preliminary tests.
With automobile manufacturers already interested in DAS technology and full-scale road tests planned, Barnes says road sign recognition systems could start to appear within five years.
He also believes the technology offers certain advantages over GPS-based driver information systems.
"GPS requires a fully up-to-date map of all the signs to be downloaded into the car at all times," he explains. "This requires the upkeep of a central database and constant download of changes. There has to be a wireless network available for the map to be maintained in the car, and constant GPS coverage.
"A vision-based system does not require new infrastructure outside the car, nor is it dependent on the reliability of outside systems. And once a camera has been incorporated into a car, there are many other possible uses, such as lane following, pedestrian detection, blind spot and parking aid technologies. Cameras on the driver can also detect driver fatigue."