Interview with Zou Lujun, general manager of EB China: Software will define future cars.

With the continuous development of technologies related to automotive electrification, automation and intelligent interconnection, popular technologies and products such as 5G technology, human-computer interaction and driverless driving are gradually growing, and the future travel mode is changing. With the development of technology, the automobile industry is undergoing rapid and profound changes.

In this era of Internet of Everything, any product may become a smart terminal, and cars are no exception. For the automobile industry, software-defined automobile has become the main theme, which makes almost all automobile enterprises face the subject of digital transformation.

Elektrobit (EB), as a global supplier of embedded interconnected software products and services, actively caters to the changes in the industry and introduces technical solutions. How should Elektrobit respond to the opportunities and challenges brought by the rapid changes in the industry? In addition, what is the position of EB in the industry, and how do you view the "software-defined car"? On these issues, we recently held the 2020? At the China Automobile Inspection and Quality Monitoring Expo, we had an exchange with Mr. Zou Lujun, general manager of EB China.

Q: EB is an embedded software company. What is the definition of embedded? In addition, what is the relationship between the current trend of separation of automobile software and hardware and embedded system?

Zou Lujun: The so-called embedded is "embedded?" "System" refers to a system on chip, usually some relatively simple 32-bit CPU or some MCU computers, with many standards. As embedded development, even if there is a chip, there will be an operating system. Different from the X86 standard of PC, it is a small system that specializes in solving some functions in a specific field, instead of giving you general function management and general calculation, it does special calculation for a certain field. Therefore, we call this kind of electronic system specially developed for a certain field embedded system.

The separation of software and hardware is aimed at the embedded market. In a highly standardized market, there is no concept of separation of software and hardware. For example, a computer PC, its software and hardware are separated. For example, if you use a computer, you install a program. Maybe you need to install Apple's application specially, but if you are an ordinary X86, all your software can be used no matter what brand of computer you download. He won't look at which chip you use. It's highly standardized. The biggest problem of embedded platform is that it is quite nonstandard compared with these highly standardized markets. Take the mobile phone as an example, the early mobile phone was also an embedded system, but the mobile phone market is very large and independent, and the mobile phone ecosystem has been established, and now the car also has this trend. However, at present, the embedded system of automobile is not standardized compared with mobile phone and home computer PC, and the difference of chips is still very big.

Q: What impact does the separation of software and hardware have on embedded software enterprises?

Zou Lujun: Actually, the separation of software and hardware was put forward by many car manufacturers, because the advantage of decoupling software and hardware is that the choice of car manufacturers can be more flexible, and he can choose software companies without much dependence on hardware. In the past, if you wanted to choose a chip scheme, you had to choose a software company supported by the chip. Because many chips support limited software, not all the functions you want to choose can be supported, so there will be all kinds of troubles and restrictions. Once the software and hardware are decoupled, the freedom of choice will increase for the car factory, so the cost will be better and more controllable, so it is a very dynamic thing for the car factory to do. For us software vendors, there will actually be a huge workload, but fortunately we are doing AUTOSAR. The core essence of AUTOSAR is the separation of software and hardware. In fact, the so-called workload of software and hardware separation is actually handled by the car factory. Now all this work has been transferred to our software company, and the workload has also been transferred to our side. We just deliver the result of the separation of software and hardware to the automobile factory. For example, when there was no separation before, you had to let the A chip cooperate with the B software. The car factory may have to pay a development fee to the software company to transplant its solution to the A chip, and the car factory will take the lead in doing this. This is an inseparable cooperation, not a technical issue, but a commercial issue. Once separated, the car factory will think that this is your own problem, and you need a software company to complete this adaptation of all the underlying hardware platforms.

Q: The current domain controller will develop into a centralized architecture in the next generation. What are the main problems of domain controllers, and why should we have a centralized architecture? What are the main challenges?

Zou Lujun: Actually, the domain controller is not too outdated. Europe has just made a domain controller, and many ECU in China are still traditional distributed ECU. First, explain why we should make it a domain controller, because we need to combine 70- 100 ecus with the same function into more than a dozen domain controllers. One reason is to consider the cost, because maybe the parts can be simplified, which can save a lot of wiring harnesses and get immediate results. Furthermore, the centralized architecture of domain controllers is mainly to further merge and save costs, because there are more than a dozen traditional domain controllers, which are now like Volkswagen HPC and three high-performance integration units. Generally speaking, in the future, Volkswagen will have three large computers to solve all electronic control problems. This is the first one.

The second is not only the consideration of cost or the reduction of wiring harness, but also that after the centralized architecture is realized, all software functions can be highly integrated on several domain controllers, and the purpose of integration is to do OTA. Without this centralized architecture, all the functions of these softwares were previously embedded in various domain controllers or a single ECU. Are these all Tier in soft mode? 1 to master, so you need to update the function, and the car factory requires Tier? 1, there will be business costs and response costs, and it is very troublesome to do one thing. And if you want to do a function, it is likely to cooperate with several different ECU manufacturers, because this kind of trouble makes this operability almost zero. So why can't the software be updated after the last car is mass-produced, or the update cost is high? The simplest example is a map. To update the map information of the car we bought five years ago, we may have to copy it to CD-ROM, which is very expensive, because the architecture and software at that time were not integrated centrally, and ECU and suppliers did it themselves. The purpose of centralization is that the responsibility of these softwares is handled by Tier? 1 Go to the car factory and make a software platform independent of hardware, so that you can talk to the software company directly. If there is a need for functional upgrade, the car factory does not need to find Tier again. 1, but directly update the business with the corresponding software vendor. For example, in the past, the update of maps had to rely on hardware manufacturers to provide ECU for car machines. Once they are made into a central integrated architecture, layers? 1 is just a hardware supplier. He can directly find his graphics provider, such as Gaud, and update the software directly without going through the second layer. 1 is for transit, so it is very necessary for this business car factory to do it by itself, because only by doing so will the value be transferred from Tier? The hand of 1 is transferred to the hand of the car factory, which is also the core purpose of software definition. Why do car manufacturers like to push software-defined cars? Is to move this value chain from the second level. When 1 supplier is transferred to itself, it is necessary to update the architecture, so as to achieve this goal and truly control the value of these car factories.

Q: Why should we shift from central structure to regional structure?

Zou Lujun: From the central building to the regional building, I understand that it is not necessarily now. One of the wire harnesses is not so much, the weight is reduced, the cost is reduced, and there are still some bottlenecks in computing power. If it is purely a central architecture, for example, many automobile manufacturers evaluated whether it is possible to control all the functions in the car and make a super-large-scale supercomputer with an HPC, but later it was found that even if it was made in theory, you might not find a chip with such strong computing power, and there is none at present. This is also the reason why Tesla does autonomous driving. The first generation used NVIDIA's chips, and the second generation was self-developed. Why does it do self-research? This is also a consideration, because its business needs have exceeded the computing power of ready-made products that can be found in the market. If it is highly integrated into a centralized architecture or even an HPC, it is very difficult to find the computing power of the chip and provide some platforms that he can use.

There is also heat dissipation. Even if such a chip platform is found, there must be a problem with heat dissipation in order to cope with such a massive computing demand. If a unified computing unit or a piece of hardware is used to do this, it can be inferred that its heat dissipation is very amazing. So, why is the area controller separated from this point of view? It may be better to separate the heat dissipation from the actual performance of the current hardware. At present, there are almost three kinds of popular HPC, one is entertainment system, the other is traditional body control, and the other is automatic driving. In fact, it can be integrated in theory, but if you want to find a very powerful chip, all the known chips on the market may not meet the requirements. There is also a heat sink. Finding such a chip is an oven. It is actually very unsafe to put the oven in the car.

Q: How is EB's global business going now? Can you share your future plans or ideas in the China market?

Zou Lujun: EB now has more than 3,000 people all over the world and has offices in 1 1 countries and regions. As you can see just now, on the road to our success, we have established deep-rooted cooperative relations with many famous automobile brands, and automobile manufacturers also trust us very much, including some traditional automobile brands, such as ABB and Ford, and have cooperated with EB for many years.

EB has always attached great importance to the China market. We set up a China office on 20 12. It's almost eight years now, and our team has been expanding. Now EB has a team of more than 100 people in China, mainly supporting some local projects in China. Of course, we don't just regard China as a sales market, but as a part of global R&D, so some foreign projects are also developed by China team. So we regard China as a market and China as a design center. Next, we will attach great importance to and expand the China market, and we will strive to have some in-depth cooperation with more local customers, including some domestic independent brands, which is why EB is willing to pay more attention to this resource. Next, the engineering team and professional knowledge will be more involved in supporting local customers.

Q: As one of the earliest promoters and initiators of AUTOSAR, how much achievements EB has made in AUTOSAR at present, what difficulties it has encountered during this period, and what is the current application of AUTOSAR?

Zou Lujun: EB is now the premium of the entire AUTOSAR institution? AUTOSAR, a member (senior member), has developed very rapidly for more than ten years since its establishment. Europe has formed a standard, and the United States is now following suit. There were few CP platforms in the early North American market, but now the transition to AP North American market has been followed up, and AUTOSAR has developed very rapidly. Especially in China, we are delighted to see that many domestic manufacturers are following the example of AUTOSAR and are launching some models and solutions supporting AUTOSAR platform. I think AUTOSAR is actually an "automobile open system architecture" from the name. Why do you want to be an AUTOSAR organization? In fact, it is to prevent each car factory from building cars behind closed doors and finally wasting some resources on software investment, so everyone forms an open standard, and then takes it out and shares it with every member, so that everyone's investment in this area can be shared equally, and everyone does not need to come to Gai Lou behind closed doors. So in this sense, the help and promotion to the whole industry is very huge. At home, we also see that after standardization, many enterprises actually get twice the result with half the effort in application development, so this piece has been developed a lot.

But it is true that AUTOSAR has also encountered some challenges. These challenges are mainly aimed at a new generation of new architecture, and the whole platform is transitioning from a classic platform to an adaptive platform. The classic platform has been clearly defined, and all specifications, standards and test scenarios have been clearly defined, which has become an undifferentiated standard that everyone can use. However, under the adaptive platform, the whole architecture has not been fully finalized. Some car manufacturers choose their own architecture, while others put forward different standards. The whole industry has not fully reached an understanding of the standards, so it is still groping, and some patience may be needed on AP. But I believe that as long as we uphold the concept of openness, this standard will definitely become a set of standards that will benefit all manufacturers in the industry. AUTOSAR does not belong to any car factory or country, but is an open standard for the whole industry, aiming at benefiting all enterprises willing to invest in software development under the open standard.

Q: It is difficult for a company to make achievements in software now. Can EB's circle of friends be shared with us, except car companies? In addition, where will the focus of cooperation with Baidu or domestic Internet companies be?

Zou Lujun: Because we focus on making car software, the ecosystem is also very important to us. In different fields, we also attach great importance to cooperation with various ecosystems. The bottom layer is mainly some chip manufacturers. For example, in the project with Volkswagen, we have very deep cooperation with Marvell and Renesas. For example, Marvell is the first company in the world to introduce an in-vehicle Ethernet solution. Including in China, we attach great importance to cooperation with the local ecosystem. For example, we have a very deep cooperation with China local chip manufacturer Xinchi Technology, including their provision of some AUTOSAR standard tools. Including us, we also have very close cooperation with some local ecosystem partners, including Baidu Apollo, which is doing autonomous driving. We are also a member of Baidu's Apollo program, and we have also contributed some EB underlying software architectures and modules to them. So one is a chip manufacturer and the other is a local ecosystem. We will also cooperate with some related manufacturers, such as some operating system companies, such as QNX, and we will have good cooperation. In some projects, we will have some joint solutions to deliver to customers.

Our cooperation with domestic ecological partners has always maintained an open mind, and if there is an opportunity, we will try our best to participate in these cooperation opportunities. As far as EB is concerned, our mission in China is to promote the domestic market to better meet some challenges brought by this new structural change. As long as we can help automobile manufacturers to speed up the landing of products and help them launch products quickly, we will explore the cooperation of all these ecosystems and we will keep an open mind.

This article comes from car home, the author of the car manufacturer, and does not represent car home's position.