One way to shrink ADAS size, power and cost

Sophisticated automotive electronics systems are moving down-market at a pretty fast clip thanks to a lot of hard work by automakers, suppliers, and in some cases, the spur of government mandates.

Advanced driver assistance systems (ADAS), for example, are likely to get a boost from regulation in the not too far distant future. The U.S. National Highway Traffic Safety Administration (NHTSA) has proposed a rear-facing camera requirement, and legislation for front-facing cameras may not be far behind. New Car Assessment Program (NCAP) ratings in Europe are focusing on active pedestrian safety, for which front-facing cameras are suited.

Regulators in the U.S. have predicted that vehicle cameras and viewing screens could cost the auto industry as much as $2.7 billion per-year or $160 to $200 per-vehicle, so cost is a significant factor in camera system design. So is size and, as always, power consumption.

“Today’s collision avoidance systems are typically enabled by digital signal processors (DSPs) or field-programmable gate arrays (FPGAs),” says Ray Cornyn, vice president of Freescale Semiconductor’s Automotive MCU Division.

Freescale is taking a different approach. Last month it announced it has licensed image cognition processing (ICP) intellectual property (IP) from CogniVue Corporation (cognivue.com) and will be the exclusive provider of CogniVue ICP technology to the automotive market. This week Freescale announced a new family of processors incorporating CogniVue technology and CogniVue announced a new generation, APEX-2™. CogniVue CEO Simon Morris said the new generation makes significant advances in vision processing performance over the previous generation, which he contends holds a 10x advantage over competitive approaches.

Preparing Recommendations

CogniVue and Freescale’s technology is based on a parallel image processing architecture that is said to enable concurrent processing of image data, for higher performance at a lower clock frequency. Pattern recognition, detection and classification algorithms extract application-specific information from a scene and interpret the image data to make decisions or take actions. Smart camera systems based on the technology are designed to detect objects and people around the vehicle, measure distance, and alert the driver of an impending collision. The sooner technology like this is widely available, the better.