Stereoscopic vision: it gives predators the ability to pounce on its prey, and lets primates leap confidently through jungle canopies. But soon it may be coming to the urban jungle as a part of your car.
Stereopsis is made possible by binocular vision, or, in the case of a vehicle, by two cameras mounted on opposite ends of the windshield and configured to focus on a single point. The two images that are perceived are then analyzed for any differences: closer objects will appear more disparate, while distant objects will seem more similar. This in turn enables the viewer, be it a human or a computer, to identify objects' size and position.
But as you might imagine, these specialized cameras are not cheap, which may explain why their debut, set for early 2013, will be on a Mercedes-Benz S-Class. The vehicle is designed to automatically apply the brakes to avoid hitting anything in the car's path, and do so more quickly than previous systems.
The system operates with the help of a radar sensor that scans for objects over 200 yards ahead, compensating for the cameras' limited 40-yard range. The radar also helps the systems in low-visibility conditions and inclimate weather. The stereoscopsis, however, allows for advanced object recognition which allows it to stop for pedestrians, but, unfortunately for Fido, prevents the car from unnecessarily swerving to avoid small animals.
As previously mentioned, similar solutions, such as Volvo's City Safety and Pedestrian Detection system, have already been employed. However, they were designed with low-speed maneuvers in mind, and thusly only work at 20 mph and slower. The Mercedes system, on the other hand, is designed for high-speed maneuvering, working at up to 45 mph.
And in systems such as Continental's Emergency Steer Assist (ESA), if there isn't time to break, the car will automatically swerve to avoid any clandestine pedestrian.
In a related article, we discussed the future possibility of self-driving cars. Several obstacles for automakers concerning safety and reliability standards were discussed, but if car manufacturers continue developing technologies such as these, hyper-aware self-driving cars may be here before we know it.
Full story at Wired