Believe it or not, imaging technology plays an important role in the cars and trucks we drive. No, this has nothing to do with the selfies you snap when you’re stuck in rush-hour traffic. Instead it involves advanced vehicle hardware.
The National Highway Traffic Safety Administration (NHTSA) will force rear-view cameras become mandatory by 2015. This is going to force automakers to include them in every vehicle they build.
But the state of this technology today is rather poor. The images most cameras provide are blurry, distorted and impossible to make out in darkness or inclement weather. But one California-based company says it can drastically improve the quality and versatility of these little imaging devices.
GEO Semiconductor has developed it calls the eWarp computing core. It sounds like something straight out of Star Trek and in some ways it really is that advanced.
John Casey, the firm’s vice president and general manager said “geometric processing is something not a lot of people are familiar with,” but the benefits of it can be striking. When applied to backup cameras this technology takes the distorted image provided by a typical fish eye-style lens and flattens it out into something the human eye can actually make sense of.
But there’s more. Casey said, “it’s not just a tool for correcting a fish eye-style lens.” With the right software it can actually pan the image displayed on the screen with minimal distortion. This might be useful while going around a corner in a congested urban area. eWarp can also provide multiple video feeds from the same sensor meaning you can get two separate images (up to eight in fact) simultaneously from one source.
Additionally this technology can readily process HD video, something that’s not really in the market today but is a trend going forward. eWarp can also handle cameras with extremely wide fields of vision, some of which can hit 185 degrees – they can actually see behind themselves! However, the images provided by these sensors are extremely distorted and requires lots of processing to make usable.
Understandably the necessary computations to achieve all of this are challenging. “Mathematically it is very complicated,” Casey said. Despite the challenges all of these calculations happen on the fly with minimal latency meaning there’s barely any time delay from what’s happening behind your SUV’s bumper and what’s shown on the screen; it’s basically a real-time stream.
Ahead To The Past
GEO’s been in the geometric processor business for about 10 years. “We’re on what you could call the fourth generation [geometric processor],” Casey said. Their technology has been used in military applications, image projectors, security monitoring and more.
Beyond these uses the company hopes to integrate with another vehicle system: head-up displays, HUD for short. These things project an image, typically on a car’s windshield, so the driver can see information right in their field of vision, stuff like speed, fuel level and even what radio station is playing.
GM held the original automotive patent for this technology and it’s something that’s still offered in many of their vehicles. But to make HUD work in a car or truck the windscreen has to be specially made. If it’s not right the reflected image will be all out of whack. Casey said this extra work can add anywhere between $100 and $300 to the cost of a windshield; in automotive terms it’s a huge expense.
But geometric processing could be a much more affordable option for HUDs. Using eWarp technology an image could be tweaked in such a way that when it’s projected onto a regular windshield it appears normal; it could also be used to project much larger images onto windscreens compared to what’s being done today.
Whether it’s applied to backup cameras or HUDs, this advanced geometric processing looks like a promising technology. But if it has any chance of catching on in the automotive space it’s got to be reasonably priced. OEMs are extremely cost sensitive. Adding another feather to its cap, Casey said their geometric processing system is very affordable since it’s little more than a computer chip.
When asked how soon eWarp technology might find its way into a car Casey said, “On the HUD side it would be sooner than the camera side.” He estimated it could show up in 18 to 24 months. As for cameras the estimated wait time is a little longer, anywhere between 24 and 30 months. These long delays are courtesy of how long it takes to develop and validate new vehicles.