The chip designer Ambarella has announced two new chips for automotive cameras and advanced driver assistance systems (ADAS) based on its CVflow architecture for artificial intelligence processing. The Santa Clara, California-based company introduced the CV22FS and CV2FS automotive camera (SoC) systems with CVflow AI processing and ASIL-B compliance to enable critical safety applications.
Ambarella will also demonstrate applications with its existing chips, as well as a robotic platform and Amazon SageMaker Neo technology to train machine learning models, at CES 2020, the big technology fair in Las Vegas this week.
The company, which was made public in 2011, started as a manufacturer of low-power chips for video cameras. But he turned that ability into computer vision experience and launched his CVflow architecture in 2018 to create low-power artificial intelligence chips. Now it has 760 employees and is competing with Intel and Nvidia, although with a focus on low-power applications.
Senya Pertsel, senior marketing director at Ambarella, said in an interview with VentureBeat that customers, such as German software manufacturer Hella Aglaia, chose CVflow due to low power consumption.
"If you look at the market, you will see that the form factor is very limited and yet it has to work in extreme temperatures," Pertsel said. “And you have to detect things like cyclists, pedestrians, etc. You have to detect over long distances … and see things like cross traffic, which requires a wide field of vision. Therefore, it is aimed at higher resolution cameras. It needs more and more processing power for complex AI algorithms, with the restriction of energy consumption. "
Both new chips are aimed at forward-facing monocular and stereoscopic ADAS cameras, as well as computer vision ECUs for Level 2+ and higher levels of autonomy for autonomous cars.
With extremely low power consumption, the CV22FS and CV2FS make it possible for car manufacturers to meet the performance requirements within the power consumption limitations of the front ADAS cameras mounted on the windshield of a single case.
Other potential applications for processors include electronic mirrors with blind spot detection (BSD), interior driver and cabin monitoring cameras, and panoramic vision (AVM) monitors with parking assistance.
The two new SoCs are the latest additions to Ambarella's successful CVC SoC family, which offers automotive companies and software development partners an open platform for differentiated and high-performance automotive systems.
ZF, a global provider of automotive systems for cars and commercial vehicles, is working with Ambarella on display and detection systems. Aaron Jefferson, vice president of ADAS product planning at ZF, said in a statement that companies are working on surround vision, electronic mirror and computer vision technologies.
Hella Aglaia, manufacturer of intelligent visual perception software, has worked with Ambarella CVflow processors for the past year. The customer chose SoC CVflow because of its ability to offer extremely high computer vision processing performance with very low power consumption, said Kay Talmi, managing director of Hella Aglaia, in a statement.
The CVflow architecture in the CV22FS and CV2FS provides computer vision processing at resolutions of 8 megapixels or more at 30 frames per second for object recognition over long distances and with high accuracy. Each of the SoCs includes a dense optical flow accelerator for simultaneous location and mapping (SLAM), as well as distance and depth estimation. Ambarella says its multi-channel high-speed sensor and image signal processing (ISP) input channel provides the necessary camera input support, even in difficult lighting conditions. CV2FS also allows advanced stereovision applications by adding a dense disparity engine.
Ambarella will show its CVflow SoC family to selected clients and partners during CES 2020. Demonstrations will include Hella Aglaia's deep learning ADAS algorithms and the Ambarella EVA (Embedded Vehicle Autonomy) autonomous vehicle prototype. Ambarella will also demonstrate a range of applications from other key partners that run on the CVflow engine.
CV22FS and CV2FS are expected to show Ambarella customers in the first half of 2020.
CES 2020 Demos
Among the CES 2020 demonstrations, Mercedes-Benz will demonstrate its load recognition and organization system (CoROS) based on CV2. A camera assistant in the cargo space automatically recognizes registered parcels using barcodes and symbols outside the parcels. This process is done in fractions of a second, replacing the manual scanning, which takes a long time and the classification of each shipment.
And Hella Aglaia will present its latest set of deep learning ADAS algorithms, which includes detection of objects of various kinds, detection of driving area limitations, depth estimation and classification of traffic lights and traffic signals. Operating with a single Ambarella CV22 CVflow SoC, this ADAS platform supports the development of forward-facing ADAS cameras in a single box.
StradVision, based in South Korea, will demonstrate its set of front-end ADAS algorithms and controller monitoring system (DMS) that run on a single CV22. Connected to an 8-megapixel front camera and an additional interior camera, this system will be installed and will work in a vehicle.
The Israel-based EyeSight controller monitoring solution (DMS) will be displayed on a system with three cameras. In this demonstration, the Ambarella CV25 will simultaneously process a driver-oriented monochrome camera and two RGBIr cameras in the cabin (each with a different field of view).
And the Israel-based Brodmann 17 ADAS solution suite will show the company's deep learning algorithms, including vehicle detection, distance estimation and real-time frontal collision warning, which also runs on a CV22 SoC.