[Introduction]Ambarella (hereinafter referred to as “Ambarella”, NASDAQ: AMBA), a semiconductor company focusing on AI visual perception, won the “Most Influential Chip Enterprise of the Year” in the first “Knowledge Award”.
On August 10-12, 2022, the second “Yanzhi Smart Car Annual Conference” hosted by Yanzhi Automotive Technology was held in Suzhou Jinke Hotel. The annual meeting lasted for 3 days, with 11 sub-forums. It received strong support from nearly 100 industry experts, attracted more than 2,000 registrations and more than 800 audiences.
On the morning of the first day of the conference, the first “Zhiding Award” was unveiled, which aims to recognize the enterprises that have made outstanding contributions to the promotion of the orderly, healthy and rapid development of the new four modernizations of smart cars. The “Zhiding Award” has been strongly supported by Academician Chen Qingquan and 60 expert judges in the industry. After layers of selection, 38 annual awards were finally selected.
At the “Yanzhi Smart Car Annual Conference”, Sun Luyi, Senior Director of Software R&D at Ambarella Semiconductor Technology (Shanghai) Co., Ltd., shared Ambarella’s “algorithm first” in his speech on “The Power of Intelligent Domain Control Chips with Large Computing Power” Strategy and chip series that fully empower automotive applications, as well as the latest progress of the latest high-power automotive domain control chip CV3 launched by Ambarella at the beginning of the year.
He pointed out: “China’s smart cars have entered a period of rapid development, and the configuration rate of intelligent assisted driving has begun to iterate rapidly. Multiple development paths coexist, resulting in a demand for diversified smart chips. The market needs high performance, high reliability, low power consumption, and high cost performance. , chips that are easy to develop software.”
He said that autonomous driving faces many problems, such as complex road scenes, complex weather and surrounding environment; smarter and more accurate algorithms are needed to meet safety requirements, and sensor failures need to be dealt with.
For domain control systems, higher-pixel image sensors, higher-performance millimeter-wave radars and lidars, better fusion algorithms, lower computing delays, and more complete functional safety designs, while supporting continuous algorithm upgrades, iterations and calculations. Force pre-embedding puts forward higher requirements for domain control chips. Among them, large computing power, especially low-latency computing, and support for various common AI algorithms, easy-to-use toolchains, low power consumption, functional safety, and technical support have become indispensable prerequisites.
He introduced that Ambarella’s latest high-power automotive domain control chip CV3 adopts a 5nm low-power advanced process, and integrates multiple sensors on a single chip to achieve centralized Al perception processing (including high-pixel vision processing, millimeter-wave radar, lidar and Ultrasonic radar), multi-sensor deep fusion, and autonomous vehicle regulation. With a complete set of mature software tools and a powerful SDK, CV3 helps car companies and Tier 1s to achieve full coverage of ADAS systems and L2+ to L4 autonomous driving products.
With the new CVFlow architecture design, the performance of CV3 running various neural network algorithms is 3~6 times that of the strongest domain control and large computing power chips in the industry, and the power consumption is less than half of theirs. These performance indicators will continue to be confirmed in demonstrations to customers.
In addition, Ambarella acquired Aoku, a 4D millimeter-wave radar algorithm company, last year. On the same 4D millimeter-wave radar hardware platform, if the Aoku radar algorithm is used, the accuracy of point cloud density and angular resolution can be greatly improved. CV3 adds hardware for processing 4D millimeter-wave radar, making the processing performance of 4D millimeter-wave radar exponentially improved.
Today, in order to accelerate the development of autonomous driving technology, a new generation of AI technology and algorithms are continuously supported, and a chip platform with large computing power and support for multi-sensor fusion has become one of the key factors in winning the competition. In the evolution of the distributed Electronic and electrical architecture to the domain centralized architecture, the intelligent domain controller chip with large computing power will play an increasingly important role. As an emerging force in the industry, Ambarella CV3 will, with the support of Ambarella China’s local engineering research and development, bring Chinese customers the best choice for high-performance, high-reliability domain control chips with high computing power.
Ambarella’s products are widely used in artificial intelligence computer vision, video image processing, video recording and other fields, including video security, advanced driver assistance systems (ADAS), electronic rearview mirrors, driving recorders, driver and cabin intelligent monitoring, Intelligent car driverless and robot applications, etc. Ambarella’s high-performance, low-power AI processors provide ultra-high-definition image processing, video compression, and powerful neural network processing to extract valuable data from high-resolution video and radar information for use in intelligent sensing, sensor fusion, and The central domain control processing system and other fields have shown their talents. For more information, visit www.ambarella.com.