A new image recognition system-on-chip (SoC) could help bring smart camera applications to future vehicles, while enabling automakers and tier-one suppliers to add their own custom features to the technology.
Known as R-Car V3M, the new SoC is targeted at applications such as automatic braking and vehicle surround view, as well as detection of lanes, vehicles, pedestrians, and traffic signs. While smart image processors have been available for those applications in the past, the V3M is said to depart from the so-called “black box” approach of the past, in which users can’t see or tailor the software code to their liking.
The R-Car V3M system-on-chip will be targeted at image recognition applications, such as automatic braking in vehicles. (Source: Renesas Electronics America)
“We wanted to give our Tier Ones and our OEM customers the ability bring their own IP [intellectual property] into the equation,” John Buzcek, senior manager for ADAS at Renesas Electronics America , which makes the new chip. “To do that, you have to have programmability within your cores, and you need to offer flexibility, as well as ease of use.”
The new SoC incorporates an ARM Cortex-A53 processor, a computer vision core, and advanced driver interfaces for Ethernet, CAN, and video, among others. Renesas says the new chip is the first standalone SoC to meet ASIL C, an automotive safety integrity level described under ISO 26262. It arrives at a time when the auto industry is preparing for a federal mandate that calls for all vehicles to offer automatic braking by 2022.
The V3M is the first product to be rolled out under the auspices of the Renesas Autonomy platform, which delivers scalable hardware, software and IP building blocks for advanced driver assistance systems (ADAS). The Autonomy platform was announced in April.
The new chip is said to be consistent with one of the major goals of Autonomy in that it focuses on openness. Openness provides a hardware and software platform, Buzcek said, while at the same time enabling users to see and understand the underlying code. They can also add their own algorithms for certain functions, such as the detection and classification of objects in front of, or around, the vehicle. That way, the automaker or tier-one supplier can have more of a hand in the autonomous car’s performance.
“There are different techniques for all those features, and we want to enable tier ones and OEMs to invest in creating their own innovation,” Buzcek told us. “At the same time, we have to provide a processor that works for those applications and supports those functions well.”
Industry analysts noted that openness is indeed a differentiator for the new Renesas chip. Up to now, competitors have typically worked with OEMs and provided a complete package, but have stopped short of supplying an open architecture that lets the automaker participate fully, they said.
“Openness lets each manufacturer do their own testing and validation, build their own software, and install policies that are particular to the way their vehicle handles a situation,” said Mike Demler, a senior analyst