Just like in Baidu some time ago? Apollo? At the ecological conference, Li Zhenyu, vice president of Baidu and general manager of intelligent driving business group, said, "? Automatic driving ability will become the decisive point of smart cars. We can also see that more and more OEMs are beginning to increase the layout of autonomous driving. However, to do autonomous driving, there are two key points: data and computing power (and the algorithm behind it). The former can be obtained by OEM, while the latter can only rely on semiconductors, and further, chip suppliers.
NVIDIA and? Mobileye？ It is recognized as two chip giants in the industry. If you want to do autonomous driving, these two companies can't escape. So, which of these two companies will become the overlord in the future chip field? This is a very interesting question.
Auxiliary driving, Mobileye? Declare oneself a king
Since? 2008? The year is over? 20 19? It's the end of the year, Mobileye Shipments exceed? 5000? Ten thousand dollars? EyeQ？ Chip; Join hands? 27? Home? OEM partners, about? 300? What kind of car? Adas? The system provides support; ? Current mass production? 1 1? Kind? In L2+ system, there are? 8? What is the basis? Mobileye？ At the same time, technology 26? Car company? 47? A cooperative project is under way.
What about this year? Mobileye？ Once again, I won several OEM orders.
This year? 1? June 2020? Meanwhile, move your eyes? The cooperation with SAIC will be realized through SAIC's subsidiaries? Mobileye？ Road network asset management? (rem? ) technology is applied to the cars produced by SAIC. Will SAIC also be equipped with a new generation of intelligent driver assistance system (ADAS)? Mobileye？ The latest visual chip? At the same time, the two sides will also conduct * * *? L3-L4？ Research cooperation level of intelligent driving project.
This year? 7? Month, Mobileye? Reach a cooperation with Ford, * * * all belong to Ford? Global product line? Develop more advanced advanced driver assistance systems, and the cooperation mainly focuses on? L 1？ And then what? L2？ Let's go
This year? 9? Month, Mobileye? Reaching a cooperation with Geely will solve its full-stack surround vision? Mobileye？ Supervision? This scheme is used for the brand-new pure electric luxury coupe concept car of Geely sub-brand? Zero? Concept? Up, the dimension of "L4" is reduced to? L2？ Use. In addition, several sub-brand products of Geely Automobile Group will be equipped with? Mobileye？ Support includes emergency automatic braking (AEB) and lane keeping system (LKA)? Adas? Function.
If nothing else, just the list of these three OEMs won this year can be made? Mobileye？ what's up EyeQ？ The chip becomes stable? At least millions of installations, with? Mobileye？ Continuously acquire and bind new customers. It has also continuously squeezed the living space of traditional suppliers.
There are two flowers, one for each table.
What time? Mobileye？ In the world? L2？ While harvesting assisted driving, Invista also began to make tactical adjustments: from exclusive use? L4？ The autonomous driving market bends down and goes deeper? L2？ Assisted driving market.
This year? 5? June, Lao Huang (founder of NVIDIA and? CEO？ Huang Renxun) moved out of his kitchen? L2？ Autopilot chip, power consumption? 5W, calculate the force? 10TOPS (for comparison, Mobileye? EyeQ4？ Power consumption? 3W, calculate the force? 2.5 times). This chip will be in? 2022? It will be officially mass-produced in the second half of 2008. By then, it will be a real confrontation between the two companies.
Autopilot, uncertain situation
In terms of advanced autonomous driving, Mobileye? Also showed his offensive: EyeQ5? Coming 20 19? Year? CES？ Go ahead, Mobileye? Reveal? EyeQ5？ Related parameters: 7nm? Craft, calculation? 24 stops, power consumption? 10kW(EyeQ4？ Counting power? 2.5TOPS, power consumption? 3W).
EyeQ5？ The chip will be available in two different versions, closed version and open version.
Some time ago, the supplier Magna announced that it was based on? EyeQ5？ Driver assistance system has been put on the market. "The industry's first system that includes a forward-looking camera and related software in one component", "The system combines Magna's expertise in electronics and cameras, and? Mobileye？ Image processing technology of system-on-a-chip (SoC) based on. This camera is second-hand? 120 800? At the same time, Magna constantly improves its world-class camera manufacturing technology to achieve the quality and output required by the global automobile platform. 」
To put it bluntly, what's the difference between this version and? EyeQ4？ Same, it is a closed black box.
The other version is an open version that supports third-party programming operations. The open version of the chip will be in BMW? iX？ It's the first time for a new car to be carried. Is this the first meeting? EyeQ？ The specific effect of opening the chip depends on the final performance.
In my opinion, the cost of this open chip is not low, which is not affordable for ordinary car companies. I'm afraid only high-end models have it? Or? With what? Mobileye？ Only closely related car companies can get this version.
In addition to supplying chips, Mobileye? More? First time? Adas? What is the role of the system supplier? It seems to provide a high-level automatic driving assistance system.
Or the cooperation with Rick mentioned just now. Mobileye？ Will its full-stack surround vision be solved? Mobileye？ Supervision? This scheme is used for the brand-new pure electric luxury coupe concept car of Geely sub-brand? Zero? Concept? Let's go
And this set of things is? Mobileye？ For what? L4？ The pure camera technology in the truly redundant sensing kit developed by the A-class autonomous vehicle (AV) contains many things:
Next year, we will see more amusement facilities? Mobileye？ EyeQ5？ Chip or? Mobileye？ Supervision? The model of the solution has landed.
The line of sight back to NVIDIA side. Although in? L2？ In the assisted driving market, NVIDIA is a bit slow, but? Are you online? On the automatic driving of L2+ and above, NVIDIA already has the meaning of grabbing food from the tiger's mouth.
Remember this picture? Lao Huang put it there? L2？ Chip sending time:
Look carefully. In this picture, why? Has the representative work of L2+ been replaced? 200 times what's up Olin? Autopilot computing platform (to some extent, it implies cooperation with Ideal and Tucki). Through this? On the L2+ platform, NVIDIA changed the ideal big customers from? Mobileye？ Even Xpeng Motors, which used supplier scheme before, fell into the arms of NVIDIA computing platform. Both companies want to gradually achieve a higher level of autonomous driving ability by embedding hardware.
The traditional giant Mercedes-Benz also chose to cooperate with NVIDIA. This year? 6? Mercedes-Benz and NVIDIA will jointly develop autonomous driving by using the latter's computing platform. Based on this platform, both parties will build a new vehicle computing architecture and jointly develop it. AI？ And self-driving car applications. Including? SAE？ L2？ Horizontal sum? L3？ Class function and automatic parking function (at most? L4？ Level) and deploy the new architecture to the next generation Mercedes-Benz cars.
Why is this happening?
If you look at it this way, you will actually find that even NVIDIA and? Mobileye？ In the final analysis, the battle for chip hegemony is actually a question of chip openness, and it is also a difference of autonomous driving strategies of different OEMs.
It can be seen that people with strong software capabilities like Ideal and Tucki will pursue more customization capabilities. This is also the reason why they gave up the supplier scheme and chose NVIDIA platform for self-research.
However, we can't ignore that a large number of OEMs actually don't have strong software capabilities? So we need a powerful supplier to help them at the perceptual level. Watching, Mobileye? The effect may be the best, even if it is a black box.
In addition, the speed of transformation of automobile manufacturers is actually slower than we expected. In the next few years, a large number of Ford, Geely and Great Wall assisted driving landing models may be used. EyeQ4？ Chip. For example:? General Cadillac? SuperCruise, do you still use the chip you are using now? Will the next generation platform use EyeQ3? EyeQ4？ Chip.
Therefore, the pace of the industry is much slower. This is also the reason why NVIDIA is in no hurry and late. 2022? Only 2008? L2？ The mass production of the chip has landed.
Again, in the next few years, assisted driving will actually occupy the mainstream position.
But I can see NVIDIA's? Olin? Open platform or give? Mobileye？ It's a lot of pressure. Is this a push? Mobileye？ Be more open, EyeQ5? The open version is a good example.
Who will be king in the future?
Having said that, NVIDIA and? Mobileye, who will be the chip overlord in the future? This is really hard to say.
Why? Because the higher the level of autonomous driving, the more different the cooperation mode may be. ? At present, we can see some clues (only for the moment):
First, choose a more open hardware platform and develop software systems, such as Ideal and Tucki;
Second, the supplier completes the big task of autonomous driving, such as perception, and the rest is independent research and development, such as Weilai;
Third, directly adopt the supplier's complete set of solutions, such as? ARCFOX？ The next car adopts Huawei's complete intelligent driving solution;
Fourth, from software to hardware, all self-developed, such as Tesla.
These changes in cooperation modes will also make host manufacturers change their chip selection.
In addition, the choice of chip platforms in the industry is actually slowly increasing, and there are companies such as Horizon and Black Sesame in China. Although the sound volume of these enterprises is relatively small, they do begin to have their own influence gradually. Like the horizon has passed? 10? Ten thousand shipping threshold.
Abroad, Qualcomm released its own? Snapdragon? Riding a horse? Platform, providing a set of software and hardware, including on-chip security system, security accelerator and autopilot software stack. At the same time, it is announced that it will cooperate with Veoneer to deliver scalable advanced driver assistance system (ADAS) and coordinated automatic driving (AD) solutions, which are expected to be in? 2024? Support the mass production of automobiles. At present, it is known that some domestic and foreign customers have been based on their? Snapdragon? Riding a horse? Develop a working platform.
At present, this market has not completely solidified, not only NVIDIA and Mobileye? There are only two, and there are still opportunities.
At the end of the article, I still have a question: will there be something similar in the field of automotive chips in the future? PC？ Market situation (Intel? Occupy? PC？ Market? 8? Into the above share), there is only one chip manufacturer?
I hope this phenomenon will never happen. If a company is dominant, the industry will be a little sad. Intel).
This article comes from car home, the author of the car manufacturer, and does not represent car home's position.