U.S.-based Alpine Inc. has developed a multiple camera system to improve visibility around a vehicle, which is additionally linked to a navigation database to add further functionality. Although the first prototypes have been developed using passenger cars, Alpine believes the technology has far greater potential applied to commercial vehicles and buses.
The system, named Top View, links four cameras—placed at the front, rear, and sides of a vehicle—to a visual display in the cabin as well as the navigation database. The data is exchanged between the mapping system and the cameras by a system known as ADAS, which carries advanced digital mapping data.
The system could carry a range of data including road intersections, curves, and gradients. In the first instance, this information could be used to warn the driver in advance that speed should be reduced as the vehicle approaches a hazard.
Garry Dyson, OE Sales & Marketing Manager with Alpine Electronics U.K., explained how it would work: “The ADAS link will give the key integration with the camera system. If you’ve got real-time camera information, it can check, for example, if the vehicle has just passed an exit on the left and then check with the map system. It all comes down to how the map is coded.
“We have already had a generation-one system about for some time in testing. Obviously, the chip technology is key for improvement, and Alpine has developed a new Top View application specific integrated circuit (ASIC) device, which has reduced the package size by 70%. The power consumption has fallen from 10 W to 1 W.”
The company also has introduced image compensation for the spherical aberration of the image, produced by small wide-angle cameras, such as those used in reversing systems. Alpine has compiled some comparative data for the degree of curvature in the image delivered by a current camera and a system with the latest ASIC-corrected image.
At 0.5 m (1.6 ft), a current system would show a distance accurate to within 6 cm (2.4 in) of the actual distance; this is reduced to 2 cm (0.8 in) for the ASIC system. At 2-m (6.6-ft) distance, the curvature of the image means it is accurate to within 40 cm (15.7 in) of the actual distance for a current system, while the ASIC-compensated image reduces that to 16 cm (6.3 in).
“A camera on its own is no good; it’s just an image,” said Dyson. “What Alpine has developed is an algorithm, which has got distance detection also built into it, just using a moving image. Around five times a second, we’re sampling the image and overlaying that on the previous image to see if there is any change. We can detect obstacles moving or fixed up to about 5 m. We’ve also got face detection which detects human faces, which would be flagged up in a box immediately.”
The face detection takes image data based on flesh tones, and there are algorithms based on eye-to-nose spacing. The system also is using upgraded cameras compared with earlier systems. These offer 1.3-megapixel definition and a wide angle lens capable of providing a 180º view, compared with as little as 100º before.
“The system is driven by two key things,” said Dyson. “One is the sensor technology—the cameras—the other is that this ASIC device has improved the integration of the devices into one chip, so the system’s a lot faster and more accurate. The algorithm we’re developing is for distance detection and that’s still in development.”
For truck and bus applications, Alpine would expect the front camera to be mounted behind the windshield, the rear camera in a similar position to current reversing cameras, and the side cameras mounted in the lower edge of the door mirror housings, where the lens would be located away from snow accumulation or spray. The ASIC device takes the four wide-angle images and merges them into a single continuous flat image of the vehicle and its surroundings.
“If there’s anything in that view, a child or a bag or any other obstacle, it will show up on that screen,” said Dyson.
That would be more applicable to bus applications, but for trucks, there are different legal implications, depending on market. In many markets, it is not permitted to show an image from a rear camera while the vehicle is moving forward. But potentially, it could address the hazard associated with a left-hand-drive vehicle driving on the left-hand side of the road or vice versa, where drivers have an enlarged blind spot because they are driving from the “wrong” side of the vehicle to see clearly at angled road intersections.
In testing to date, Alpine has used an interior rearview mirror with a row of white LEDs in the lower edge, which will flash to warn of potential hazards ahead.
There are several further potential ADAS developments. First, Alpine has already conducted tests to show potential hazards ahead. “Imagine you are driving and 500 yards ahead a car has come to an intersection. The system will recognize that car as a potential hazard, put a box round it, and flash it up in red,” Dyson explained.
Developments could also see the map data used actively. If, for instance, the driver has failed to heed a warning to slow down, the system could automatically apply the brakes to slow the vehicle to a speed calculated to be safe for that vehicle. The next stage in developing ADAS systems could see wireless transmission of data between vehicles. If a vehicle had been involved in an accident, that information could be automatically transmitted from vehicle to vehicle following on the same road, using wireless links between them.
Ultimately, Alpine believes that door mirrors could be replaced with camera-based systems, reducing the aerodynamic drag caused by large mirror housings.