The software scans the video pictures and detects road signs by recognising their symmetrical shapes: rectangles, diamonds, octagons or circles. Once a sign is detected, the image is compared to a list of signs stored in the computer’s memory. If it recognises a stop sign, the computer checks if the car is slowing down.
The computer uses a commercial package called FaceLab to analyse images from the stereoscopic cameras and work out where the driver is looking. If the driver appears not to have seen a sign, and the car’s speed does not change, an alert is issued, says Nick Barnes, one of the developers at NICTA.
Most such warnings wouldn't be popular with speeders. But an adjustment could be added that lets you specify how far above the speed limit you want to go before being notified. Plus, an "Off" switch would be really helpful - especially in an emergency if you are in a mad dash to take someone to a hospital emergency ward.
Radar-based adaptive cruise control with automated speed adjustment to avoid collision first debuted on passenger cars in 1998 in an S-class Mercedes. or possibly first in a Toyota in 1997 for Japan's domestic market. More recently some Spanish researchers have demonstrated optically-based adaptive obstacle avoidance.
A camera-based cruise control system that automatically slows a car down to avoid potential accidents has been developed by Spanish researchers.
The prototype system - which uses a single dashboard camera to monitor traffic ahead - has been installed and tested by Miguel Ángel Sotelo and colleagues at the University of Alcalá, Spain.
An optical system has the advantage that it could be combined with the first reported system above to read road signs and do other image processing to detect dangers that a radar-based system might not be able to manage to pick up (e.g. a deer running out of the cover of woods close to the road). But image processing is probably algorithmically more difficult and probably also requires greater computer processing power.
A cruise control that could read existing signs could detect when the speed limit changed and automatically adjusted the car speed up or down to the new limit. Further into the future electronic bar codes that are either optical or electro-magnetic could be embedded into roadways to be read automatically by passing cars.
But the problem with all these computer computer-assisted driving gadgets is that they do not yet eliminate the need for drivers to pay attention. What we really need are cars that can drive themselves. The big savings in time will come when drivers no longer have to pay attention. Sometimes active driving can be fun and even pleasantly distracting from the rest of life. But lots of hours spent behind the wheel are just wasted time that could be used to perform various tasks. Once high speed cellular modems are installed along most roadways one could, for example, go shopping on the internet or read internet news sites while being whisked home by a robotic car. Or one could carry on intellectually more demanding phone conversations for one's job without the distraction putting at risk the lives of everyone else on the road. Or one could do more sight-seeing
I see these smaller incremental improvements in road condition sensing, automatic cruise control adjustment, and the like as helpful steps that will allow the testing and maturing of various technological components that will gradually allow computers to take over more of the work of driving. The first real step toward total automation will probably occur on special lanes on highways. The more complex environment presented by city and residential neighborhood driving will likely remaini in the hands of humans for one or two decades longer.
|Share |||Randall Parker, 2004 October 03 07:03 PM Travel Tech|