Core Intelligent Driving -- A collection of the views of famous experts in production, education, research and enterprises, a comprehensive analysis of THE opportunities and challenges of AI chips, third-generation semiconductors and other opportunities and challenges in the era of "great deformation" of automobiles!
Jiwei Network reported that in 2021, the mainland L2 level of intelligent driving has achieved large-scale commercial application. At the "2022 China Electric Vehicle 100" forum held on March 25-27, leaders of relevant ministries and commissions of the state clearly pointed out that in 2021, the proportion of domestic passenger car L2 level intelligent driving will reach 22.2%, while in 2020, this proportion will be 15%. It can be said that behind the sharp increase in the penetration rate of L2 in the mainland is the boost of the new energy vehicle market. In 2021, the penetration rate of mainland new energy L2 passenger cars far exceeds that of fuel vehicles, and it is close to 38%.
Source: 2022 China Electric Vehicle 100 People Will Give a Speech
Large computing power is the definite trend of autonomous driving chips
As an important part of the decision-making layer of the intelligent driving system, the automatic driving chip is the key hardware support for the realization of intelligent/autonomous driving. Because the core of intelligence is data calculation, the basis of data calculation is the chip, cameras, radar, positioning navigation, car communications and other components will form a large amount of data, these data are calculated by the car chip.
Large computing power is regarded as one of the definite trends of autonomous driving chips. With the increase in the level of autonomous driving, the larger the amount of data it needs to process, the higher the demand for the computing power of the chip. According to Intel's calculations, in the era of fully autonomous driving, each car generates up to 4,000 GB of data per day. According to the horizon data disclosure, every time the automatic driving level is increased by one level, the required chip computing power will show a dozen times increase, and the computing power demand for L2 automatic driving only requires 2-2.5TOPS, but the L3 level automatic driving computing power demand needs to reach 20-30TOPS, to L4 level needs to be more than 200TOPS, and the L5 level computing power demand is more than 2000TOPS.
At the "2022 China Electric Vehicle 100" forum, Wang Ping, CEO of Cambrian Xingge, said, "In the L3/L4 era, the amount of data has increased significantly, and the algorithm has become more complex, so it needs a chip with large computing power to meet the demand, in addition, the development of OTA also needs a universal and open software platform to support." ”
Shan Jizhang, founder and CEO of Black Sesame Intelligence, also pointed out that from a technical point of view, the real breakthrough from L2 to L3 will be a relatively long process, which involves software, hardware, data and other technologies with the continuous upgrading of the automatic driving system, the realization of future intelligent cars and automatic driving needs to break through the bottleneck of computing power, and the use of large computing chips is the key to obtaining a breakthrough.
Domestic autonomous driving chip manufacturers have emerged
As far as the current market pattern of autopilot chips is concerned, in addition to Tesla's self-developed FSD chips, most of the international car companies and domestic car companies that mass-produced auxiliary/autonomous driving systems are currently relying on chips such as Mobileye and NVIDIA. Of course, because these companies have a first-mover advantage in the field of automatic driving, it is understandable that domestic car companies take the lead in choosing chips with better performance, but the safety of the industrial chain also needs to be paid attention to.
The lack of cores in automobiles since the end of 2020 has made the industry realize that in the long run, the importance and necessity of localized chip supply will not only empower the landing of China's automatic driving technology, but also help build a stable and reliable supply chain, and domestic autonomous driving chip companies will become the "backbone".
For example, at this year's national two sessions, Chen Hong, chairman of SAIC Motor Group, suggested actively promoting the localization of vehicle regulation-level and large computing chip, and the state took the lead in setting up special funds to encourage chip companies and automobile companies to participate together and accelerate the formation of domestic large computing chip research and development, manufacturing and application capabilities.
In addition, we have also found that some Chinese manufacturers have gradually emerged in the autopilot chip market, and their chips have been put on the car or are about to be put on the car, for example, the ideal car uses a small amount of horizon chip Journey 3, and the EXTREME FOX Alpha S Huawei HI version of BAIC BJEV will use Huawei's chip. At the same time, there are more and more Chinese chip manufacturers entering this track, and there are mass-produced products released this year and next. At the "2022 China Electric Vehicle 100" forum, companies such as Black Sesame, Cambrian Xingge, and Horizon also shared the latest progress of self-developed auto/intelligent driving chips.
Shan Jizhang revealed, "2022 will be the mass production year of large-scale car specification chips, Huashan No. 2 A1000 series chips are scheduled to start mass production this year, will become the largest domestic production of computing power, the strongest performance of the automatic driving chip, at the same time it will also become the first mass production in line with car regulations, single chip support for the integration of domain controllers of the domestic chip platform." "At present, Black Sesame Intelligence has formed a complete solution at the L1 to L3 levels, and is constantly expanding the scale of partners, and has cooperated with nearly 70 partners.
Wang Ping said that in order to meet the needs of the smart car market, Cambrian Xingge will launch products that fully cover different levels of intelligent driving chips. Cambrian Xingge for the L2+ level market SD5223 chip, belongs to the integrated solution of the berth, using advanced technology, the maximum computing power of 16 TOPS, a single SoC can realize the function of the berth integration, and can use natural heat dissipation, to promote the automatic driving system to 8-100,000 yuan of entry-level model coverage, and this chip will be released in 2022. And for the L4 level of automatic driving solutions - SD5226 series chips, in terms of artificial intelligence computing power will be further increased to more than 400 TOPS, CPU maximum computing power of more than 300K DMIPS, the use of 7nm process, independent safety island design, the first to provide L4 level automatic driving solutions based on a single SoC, and the highlight of this chip will be to support the car end training, support the car side of the self-learning architecture, This solution is also scheduled to be officially released in 2023.
In addition, Horizon's latest product, The Journey 5 chip, is also scheduled to be mass-produced in the second half of 2022, with a maximum AI computing power of 128TOPS per chip, using a 16nm process and supporting 16 cameras, which can meet the mass production needs of high-level autonomous driving in car manufacturers. At the beginning of 2021, Horizon founder Yu Kai revealed that the planning computing power of the Zhengcheng 6 chip is 400TOPS, which will use the vehicle-grade 7nm process, and it is expected that engineering samples will be launched in 2023 and mass production will be achieved in 2024.
The mass production of large-scale car specification chips still needs to break through the bottleneck
Globally, there are signs that the development of autonomous driving will be on the fast track. In particular, the regulatory level that directly affects the actual landing of automatic driving has made a major breakthrough, And the regulatory authorities such as Japan and Germany have officially opened the door to policies for some advanced autonomous vehicles, and the United States has also said that it is expected to introduce important policies for automatic driving in the next few years, and China is also vigorously advocating.
In terms of car companies, in 2021, L3-class mass production models have been on the road in some countries and regions. Honda began selling the world's first high-end sedan, the Legend, equipped with an L3 self-driving system, in March last year, and received permission from the Japanese government to go on the road.
Following Honda, in December 2021, Mercedes-Benz became the first automaker in the world to receive international certification for conditional autonomous driving (L3 level) systems. Moreover, Mercedes-Benz recently announced that when a Mercedes-Benz car equipped with Drive Pilot opens the vehicle's advanced driver assistance system, the driver is no longer legally responsible for the operation of the car. In the event of a car accident, Mercedes-Benz will be liable. Pilot driving has been approved for use on all highways in Germany, and Mercedes-Benz hopes to roll out the self-driving system in the United States by the end of 2022.
According to Statista, by 2025, the market share of L2+L3 production models is expected to exceed 50%. Now all car companies are working hard to develop and promote the mass production of higher-level autonomous vehicle models.
The industry has entered a period of rapid development, but it must also be faced squarely, the development of domestic autonomous driving chips still faces many challenges, and the mass production of vehicle specification chips, especially large-scale car rules chips, is a difficult process, which needs to go through the necessary stages and investment.
Shan Jizhang said that at present, there are still some key technologies to be broken through from the mass production of large-power car specification chips, such as advanced packaging technology and independent IP technology in chip technology; high-computing chip system architecture needs to be broken, and has a high-security tool chain, high-security operating system and information security; in terms of vehicle regulation certification, a series of core technologies such as functional safety process, functional safety product certification, vehicle regulation reliability certification, and ASPICE software certification need to be broken one by one.
For example, the Black Sesame Intelligent Huashan series of autonomous driving chips, starting from the completion of product definition, tape-out, packaging and testing, vehicle regulation certification and algorithm tool chain ready, functional safety certification to end-customer verification, mass production on the car, has experienced more than 3 years.
In the face of these problems, Wang Ping also appealed: "From the perspective of car companies, we hope that car companies can give domestic chip companies more opportunities, through joint development projects, traction domestic SoCs to become SoCs that are more in line with the needs of car companies, and use more localization chips to improve supply chain security." At the same time, it supports and guides the construction of the ecosystem, and encourages strong cooperation between domestic chip companies, algorithm companies, Tier 1 and other enterprises. ”
(Proofreading/Jimmy)