Ambarella launches new automotive vision chips

Ambarella is unveiling a new automotive camera chip, the Ambarella CV22AQ, to provide computer vision for deep neural network (DNN) processing in advanced driver assistance programs in cars.

Santa Clara, California-based Ambarella is in full swing of converting its position in camera sensors and processors into a foothold in computer vision and artificial intelligence chips for driver-assistance cameras, electronic mirrors, in-car cameras, and parking assistance technology. The company will debut the new tech at the CES 2019 trade show in Las Vegas in January.

The system-on-chip (SoC) provides the performance necessary to exceed New Car Assessment Program (NCAP) requirements for applications such as lane keeping, automatic emergency braking (AEB), intelligent headlight control, and speed assistance functions.

The chip is fabricated in an advanced 10-nanometer manufacturing process, where the width between circuits is 10 billionths of a meter. It has low power consumption and fits in small devices such as windshield‐mounted forward Advanced Driver Assistance Systems (ADAS) cameras.

“This is a mainstream automotive product focused on forward-facing driver assistance cameras and driver monitoring,” said Chris Day, chief marketing officer at Ambarella, in an interview with VentureBeat. “These days, we are focused on security and surveillance cameras, as well as automotive.”

In another announcement, Ambarella is teaming up with German automotive supplier Hella Aglaia on a new generation of ADAS products. The platform is based on Ambarella’s CV22AQ CVflow computer vision processor, which offers a powerful Image Signal Processor (ISP) and massive Artificial Intelligence (AI) computing performance with extremely low power consumption, typically below 2.5 watts.

Combined with Hella Aglaia’s production‐proven software, the platform enables forward‐facing, single‐box ADAS cameras with advanced AI features and accurate image content recognition, requirements for safety‐critical applications.

Last year at CES, Ambarella launched two new chips to bring advanced vision processing to next-generation cars, drones, and security cameras. All of the Ambarella chips are based on the CVflow architecture, which is optimized for computer vision algorithms such as stereo processing and deep neural networks, said Day.

At CES 2019, Day said, “I expect we’ll see a lot more technology focused on driver monitoring. We certainly have a lot of requests for those systems.”

Ambarella has done very well in the security and surveillance markets. But it is a newbie challenger in automotive, competing with rivals such as in-house chip divisions at car makers and big rivals such as Intel and Nvidia. Huawei’s High Silicon division is also a main competitor in security and surveillance.

“We have a lot of people working with us behind the scenes in automotive, though it moves at a slightly different speed,” Day said.

The CV22AQ’s CVflow architecture provides computer vision processing in 8‐Megapixel resolution at 30 frames per second, to enable object recognition over long distances and with high accuracy.

Ambarella said the CV22AQ supports multiple image sensor inputs for multi‐FOV (field of view) cameras and can also create multiple digital FOVs using a single high‐resolution image sensor to reduce system cost. It enables DNNs for object detection, classification (i.e. of pedestrians, vehicles, traffic signs, and traffic lights), tracking, as well as high‐resolution semantic segmentation for applications such as free space detection.

The CV22AQ’s high‐performance Image Signal Processor (ISP) provides imaging in low‐light conditions while high dynamic range (HDR) processing extracts maximum image detail in high‐contrast scenes, further enhancing the computer vision capabilities of the chip. It includes efficient 8‐megapixel encoding in both AVC and HEVC video formats, allowing customers to add video recording and streaming capabilities to their automotive cameras. The SoC’s advanced cyber security features, which include secure boot, TrustZone and I/O virtualization, enable over‐the‐air updates (OTA) and also protect against hacking.

The CV22AQ is currently sampling to leading top-tier customers and tier-two algorithm providers. Chip samples with ASIL‐B support are targeted to be available in 2019.