The little virus is too rampant. Successfully running a conference now is like winning a big prize.
The first Jiwei Automotive Semiconductor Ecological Summit was postponed twice due to the epidemic, and was finally completed on December 24, 2021 in Pudong, Shanghai. Nothing more than three. Lao Ji believes that ecology is very important.
Dai Hui, a veteran of science and technology, has the biggest feeling that after the advent of the intelligent era, China's automotive software and the world's advanced level are basically synchronized, which can help the automobile industry expand sales and drive the development of the chip industry. There is also a "soul" in the core, which is the operation software, middleware (such as SOA service-oriented architecture) and a wide range of application software.
First, the era of "software-defined cars" has arrived!
The term "software-defined cars" has been buzzing! It is expected that the size of China's software-defined car market will exceed $82.3 billion by 2030.
Picture note: Mei Jinren of SAIC Zero Beam Technology introduces the era of "software-defined cars"
Photo note: ArcherMind Ye Liwei introduced smart car software
The era of intelligence has arrived, which is a technological revolution, and China's automobile industry has won a valuable opportunity to overtake in curves. China's local automotive software industry has contributed to it.
It is interesting to say that Dai Hui, a communications veteran, studied history and found that mobile phones have a great impact on the automotive industry.
Since 2006, the rise of the cottage machine era, MediaTek and Spreadtrum have promoted TURNKEY SOLUTION to provide a package of software and hardware solutions, and the domestic mobile phone operating system software industry has also been followed. After the rise of 3G/4G, Qualcomm made full use of China's mature mobile phone underlying software industry, engaged in QRD (Qualcomm Reference Design), and also provided a package of mobile phone solutions.
Qualcomm began to cross-border into the field of automotive intelligent cockpits in 2014, and in January 2014 Qualcomm launched its first-generation automotive intelligent cockpit platform Snapdragon 602A. On the market at the time, cockpit electronics was more fragmented. Qualcomm adopted the successful experience of the mobile phone industry, chose to cooperate with china's software industry strategically, and provided a package of software and hardware solutions for smart cockpits, which greatly promoted the development of the smart cockpit industry. The local automotive software industry has developed rapidly together and gone global. ArcherMind Technology, which participated in this summit, is the representative company of China's mobile phone and automotive underlying software, witnessing this era.
Qualcomm Sun Gang believes that the strong rise of Chinese automakers actually has similarities with the rise of mobile phones. At present, the evolution of new technologies is getting faster and faster, and Chinese automakers are faster than European and American automakers. The Chinese market has a distinctive feature, that is, there are many car manufacturers, many demands, rapid iteration, such as in automotive communications, 4G technology from 2012 to 2-3 years after the launch of the car, there are only two car manufacturers using this technology; and when the 5G is launched two or three years, it has been in more than 18 car factories, most of which are Chinese car manufacturers.
Photo note: Qualcomm Sun Gang introduced the automotive semiconductor strategy
The next 30 years will be the era of the Internet of Everything. In this era, chip technology based on the original communication and computing field (so-called ICT technology) can derive many products, and the car is one of the market segments. The development history of ARM, Qualcomm, Huawei, etc. is like this.
Figure note: ARM develops NPU (Neural Network Unit) in mainland China
Domestic automotive SoCs (System-on-Chips) are actively catching up and paying attention to cooperation with local software ecosystems. Qiu Yujing, CEO of Xinchi, once introduced that for SoC chips, in order to succeed, it is far from enough to make chips, and it is also necessary to build a software and hardware ecology. Taking software as an example, the first layer around the chip is the operating system, the extension is the middleware, and then the extension is the algorithm, such as voice algorithm, navigation algorithm, DMS algorithm, ADAS automatic parking algorithm, etc. Chip companies must adapt these details, customers directly use, you can quickly achieve mass production.
Artificial intelligence software in the car is another field, the earliest to do is MOBILE-EYE (now belonging to INTEL), and the successful strategy is software and chip coupling (now open). AI software companies such as Zhongmu Technology and Moshi Intelligent Industry, which participated in the summit, used local massive data deep learning to provide software such as DMS (driver monitoring) and ADAS (high-end assisted driving) mentioned above, and were in line with the current mainstream software and hardware separation strategy.
Picture note: Black Sesame Yang Yuxin said that the success of the chip requires a complete ecology, and Black Sesame adopts a software and hardware separation strategy
Chen Yunrong, who has worked in a mobile phone design company for a long time, believes that this is the same as mediatek/Spreadtrum's cottage mobile phone SoC in that year, and the local automotive SoC is also expected to cooperate with domestic advanced software capabilities and break through successfully.
At the Automotive Semiconductor Ecosystem Summit, several companies invariably showed a similar picture of the evolution of in-vehicle electronic and electrical architectures. ZTE is a traditional ICT (information and communication technology) manufacturer, which entered the field of mobile phones and internet of vehicles early, and has the same concept.
Figure Note: The evolution of electronic and electrical architecture narrated by Zhang Xiaoyu of ZTE
Conventional cars use distributed electrical and electronic architectures with dozens to hundreds of different ECUs (electronic control units). After the advent of the era of intelligent networking, domain-focused or cross-centralized electrical and electronic architectures are accelerating, and will eventually evolve to centralized electrical and electronic architectures based on central computers. Software has always been an indispensable key link in this evolution.
Just like the trend of platformization at the end of the past few decades, domestic and foreign auto giants have launched new vehicle platforms and a new generation of electronic and electrical (E/E) architectures, simplifying the hardware of traditional vehicle platforms, and emphasizing the use of software to create differentiated advantages, from the past around the "chassis + power system" to "electronics + software" as the center and adapt to different chassis and power systems.
In the future automotive architecture, the application of domain controllers may enter the stage of central computing units, and high computing power becomes the foundation. Software is increasingly becoming the center of a new round of market battles in the automotive industry chain.
The smart car of the future is expected to have multiple domains, such as intelligent cockpit domain, intelligent driving domain, body control/power domain, etc. As the center of future automotive computing decision-making, the realization of domain controllers (DCU) functions mainly depends on the organic combination of multi-level software and hardware such as master chips, software operating systems, middleware, and application algorithms. In the future transformation of the entire automotive electronic and electrical architecture, the complexity and difficulty of the underlying software will increase dramatically.
The OTA business model is a big boost for software-defined cars. Jingluo Semiconductor Zheng Lianqun pointed out that Tesla's newly updated tire wear detection function as an example, behind the small functional upgrade is a huge technical and business model change, last year Tesla through the OTA model to obtain revenue of more than 1 billion US dollars. In the future, the automotive E/E architecture will become more and more centralized, and multiple SoC chips will be responsible for the control and function of different areas, and will be connected through in-vehicle Ethernet, and it is expected that there will be more than 100 connection nodes for high-end models.
According to Jingluo Zheng Qunhui, new software ideas such as OTA and SOA require numerous Ethernet connections
Many international TIER 1 (Tier 1 suppliers) have set up research institutes in the mainland and invested in domestic enterprises, such as Bosch, an important reason is the rapid development of smart cars and intelligent ecosystems in the mainland.
Shi Shuai, a partner at Roland Berger, believes that when it comes to the current automotive industry, it is impossible to avoid software-defined cars, and there are two important driving factors behind software-defined cars.
First, the integration complexity of in-vehicle software has been greatly improved. Areas of software-driven innovation include connected vehicle capabilities, autonomous driving, and electrification of powertrains.
The second factor is the rapid innovation of semiconductor and software technologies, such as multi-core SoCs, AI chips, high-performance in-vehicle Ethernet networks, more service-oriented software architecture and function implementation, and cloud integration and network security technologies, etc., which will lead to the development of centralized computing platforms and end-to-end software platforms in the future, which will reduce integration complexity and cost, and enable third-party services to be commercialized.
Overall, driven by the technology of software-defined vehicles, the electrical and electronic architecture will lead to new vehicle architectures and integrations, while OEMs, including component suppliers, will build software platforms that support the service ecosystem in the future.
Second, the traditional automotive electronic controller (ECU) software has a long history
Cars are the crystallization of human civilization. Automotive electronics have been developed by the world's brightest minds over the past few decades, learning the lessons of many tragic traffic accidents.
Traditional cars are discrete E/E (electrical and electronic) architectures, and a vehicle has evolved from several ECUs to dozens or even more than 100 ECUs (electronic control units). MCU (Microprocessing Unit) To connect various sensors, it is important to program the MCU via assembly language or high-level language, software and algorithms.
Now many entry-level family cars in China are also equipped with AUTOmotive electronic safety equipment such as ABS, ESP, AEB, tire pressure monitoring, etc., saving many lives. Tier 1 (Tier 1 supplier, benchmarked by Bosch) in the West is a benchmark company and also has R&D and production bases in China.
Mei Jinren, a zero-beam software technology owned by SAIC, recalled that before 2000, there were very few electronic components in domestic cars, mainly radios, etc., which was a big gap compared with foreign automobile industries. In 2004-05, the domestic automobile industry began to introduce engine EFI, automatic transmission, brake system and other technologies. In order to solve the integration of powertrain systems and chassis systems, foreign distributed electronic and electrical (E/E) architectures have been introduced, and there are only four or five electronic controllers in a car, or 8-digit machines. Later, the number of ECUs also gradually increased, and the MCU used in addition to the 8-bit machine also had 16-bit and 32-bit MCUs, until today.
I remember at the beginning of this century, when I was learning to drive, the teacher was still telling us to brake, when I bought the first car (Beverly Wilshire) in 2006, there was an ABS anti-lock braking system, and the emergency brake was a foot in the end! When I bought my second car in 2009, I hadn't heard the word ESP (Body Stability System) on the market. This ECU was developed by Bosch in 10 years, and is an active driving safety measure that avoids skidding to a large extent.
ECUs are the foundation of automotive electronics and will be present for a long time in the areas of body, chassis and power.
The proportion of new energy vehicles is getting higher and higher in new cars, reaching about 20% in 2021, compared with only 5% in 2019. Compared with traditional fuel vehicles, new energy vehicles will use more electronic control unit ECUs.
Traditional electric vehicle manufacturers are developing their own "three electric" (electric drive, electronic control, battery management) of the overall electrical and electronic architecture, which contains a lot of ECUs. Liang Haiqiang of BAIC New Energy Engineering Research Institute made a brief introduction at the summit.
China has begun the research and development of ECU software. For example, in 2008, Fujitsu Nanda Research Institute began research on diesel electromechanical injection systems. In 2013, Fujitsu Semiconductor's MCU business was sold and transferred to Infineon. Zhou Zhihua and others who worked here founded Nanjing Meijun Electronics in 2014 to start the development of ECUs, the core is the underlying software, and adapted to the three major MCU systems of Renesas, NXP, and Infineon, and are also verifying domestic MCUs. Its role is TIER 2, the direct customer is mainly TIER 1, and the final customer is many domestic car manufacturers. The following figure is a TIER 1 electronic water pump, the controller inside is developed by Meijun Electronics, using NXP's 16-bit MCU, one of the core jobs is to write software. Domestic depot iteration speed is very fast, local companies can quickly modify the software to respond.
Figure Note: Controller for Electronic Pump (Meijun Electronics)
Under the background of "lack of core", car manufacturers are actively verifying domestic chips out of the need to "guarantee supply". This is a once-in-a-lifetime opportunity, and domestic automotive chips have almost started from 0 and quickly entered the market verification stage.
Some places do not have high safety level requirements, such as windows, sunroofs, seats, headlights, trunks, wipers, air conditioners, atmosphere lights, etc., whether it is software or chips, the country is actively importing.
In the body, chassis and power areas, the safety level is very demanding, which is closely related to the safety of the driver and is mainly used as "control". The chips (MCUs and sensors) used are mainly from the West, using mature processes, although the supply is very tight, but as long as the price is high enough, the spot market can still buy at a high price.
Regarding the programming of MCU, Dai Hui, a technology veteran, wants to blow the bull. In the early 90s, I learned assembly language programming for 8-bit microcontrollers (that is, MCUs) and 16-bit computers. Microcontrollers were originally widely used in consumer electronics and industry (such as Huawei Xu Wenwei's program-controlled switch development), and used in more demanding automotive fields.
Third, SAIC Zero Beam: Full stack solution introduces SOA middleware
SAIC Zero Beam cooperated with SAIC Motor to develop the first and third generation of electronic and electrical (E/E) architectures, accelerating its further integration of electronic architectures, in order to transfer more software development to OEMs (that is, OEMs or OEMs), so that the OEM iteration speed was further accelerated, which is also in line with market demand, because the previous development cycle of a model was basically three years, and now it is basically controlled in about 18 months. The upcoming Zhiji car has been adopted.
Among the speeches of Zero Beam Mei Jinshan, I am most interested in the "SOA Cloud Pipe End Integration Software Platform".
SOA is backward compatible with heterogeneous OS (such as Android, LINUX, QNX), provides application APIs (application programming interfaces) upwards, provides software development IDE environments, realizes application mobility and pluggability, continues to create a car-specific application ecology, and ensures that future intelligent networked vehicles can easily implement OTA and other functions. For end users who are car owners like us, we can flexibly combine services and software to achieve subscriptionable services to meet the needs of thousands of users.
Make up the lesson. SOA (Service-Oriented Architecture) means service-oriented architecture, which is a software middleware, but also a software design idea, which has been developed and applied in the IT field for many years. Through standardized service interfaces, loosely coupled service mechanisms and scalable service features, combined with the centralized electrical and electronic architecture with the high-performance computing platform "domain controller" as the core in the future, it will become the technical foundation of "software-defined vehicles", "technology-driven innovation" and "data-driven services" in the future automotive field. The figure below is the multi-chip architecture I found for Volkswagen's on-board central computer, which uses an SOA software architecture.
Figure note: Volkswagen's architectural SOA implementation in VCC
4. ArcherMind Technology: Provide services for underlying software such as operating systems and middleware
The chip computing power of the mobile phone is important, but the underlying software such as the operating system is also very important, which we all know, and the two are inseparable. MediaTek's package solutions for cottage chips have laid the foundation for successful experience.
Since the beginning of this century, ArcherMind Technology has actively participated in the underlying software business of TURNKEY SOLUTION (package solution) promoted by MediaTek, Qualcomm, Spreadtrum, HiSilicon, etc., and has entered the field of automotive intelligent cockpits. The smart cockpit is an extension of the smartphone and is a new generation of ways to connect and interact.
ArcherMind Technology is positioned as a software developer and service provider of intelligent networked vehicles, deeply cultivating various operating systems and middleware, solving the most complex software problems in the process of mass production of automotive OEMs, and allowing OEMs and Tier1 to focus on the optimization of the functions of the car itself and the user experience. The operating system and middleware are mainly responsible for the reasonable allocation of hardware resources to ensure the orderly progress of various intelligent functions.
ArcherMind Will use its own software advantages to connect the underlying middleware and operating system to help automotive OEMs significantly shorten the development cycle and save development costs.
On the operating system, ArcherMind has been working on Android, QNX, AliOS, Linux and other OSes for many years, and can also support Hongmeng OS.
Caption: ArcherMind Ye Liwei describes the smart cockpit business layout of one of the product lines
Dai Hui, a veteran of science and technology, sighed that in the past, only one operating system was run on a processor, and now after a processor passes the virtualization layer, it can run multiple different operating systems at the same time, such as QNX (the world's largest automotive real-time operating system RTOS), LINUX, Android, etc. In January 2016, on the way to CES in Las Vegas, my neighbor was QNX Greater China GM Zhang Renyan (now the public asks the CEO). Just got the news, Xiong Xinpu (jianghu people call Xiong Da) founded RT-THREAD has also entered the automotive RTOS ranks. Jinghua Kexun is also one of the earliest virtualization technology enterprises in China.
Photo note: The author Dai Hui is outside QNX headquarters in Ottawa
In terms of middleware, ArcherMind Technology provides AUTOSAR software middleware that meets the requirements of THEA architecture based on the mainstream AUTOSAR protocol stack, so that OEM companies are no longer trapped in the research and development and upgrade maintenance of chips and underlying AUTOSAR software.
ArcherMind is one step ahead in the automotive software ecology, and has effectively promoted the development of local chip SOC companies, and the cooperation between the two sides has even advanced to the definition of SPEC, which is very important in the early stage of chip design.
ArcherMind and BMW have established a joint venture to customize localized software, such as adapting to Hongmeng and ALIAS, and connecting with local car ecosystems.
Companies such as Zhongke Chuangda, Neusoft Ruichi, and Runxin Micro also provide similar services.
V. ZTE: Launched the overall framework of goldenSAR platform based on core ICT capabilities
Cars have traditionally been mechanical, and IT (information technology) and CT (communication) are constantly infiltrating. Qualcomm, Huawei and ZTE are all companies that have entered the automotive sector from communications and information technology.
ZTE has strong basic capabilities in many aspects of the ICT field, and ZTE has also entered the fields of Internet of Vehicles, automobiles, and batteries very early. ZTE integrated internal resources early last year to establish an automotive electronics business product line and cooperated with OEMs such as SAIC and FAW.
At the summit, ZTE Zhang Xiaoyu focused on the overall framework of the Automotive Electronics GoldenSAR platform.
The GoldenSAR platform adopts a layered architecture design idea, including the basic software platform AUTOSR and SOA Framework Middleware; GoldenSAR is a layered, open, and extensible platform that supports OEM and Tier1 collaborative development.
Picture note: ZTE Zhang Xiaoyu introduced one of the products, GoldenSAR
Artificial intelligence: advanced driver assistance system based on deep learning
Baidu Apollo, a number of robotaxi companies, Huawei MDC and other demonstrations are passenger car driverless. However, due to various factors, driverless driving (L4 or L5) will take some time. The new domestic car-making forces are closely following Teqila, and have also launched a function similar to "assisted automatic driving", saying that they are L2.99 (still need driver control).
Interestingly, a number of new startups have emerged in the market to provide high-end assisted driving technology to support advanced driver assistance (ADAS) functions that are implemented to varying degrees from traditional passenger cars to commercial vehicles and tool vehicles.
Photo Note: Tang Zhong, a technology company, introduced assisted driving and autonomous driving business
Zhongmu Technology is a supplier of autonomous driving and advanced driver assistance system (ADAS) products and technologies, based on parking algorithm capabilities, millimeter wave radar, superwave radar, cameras and complete system design capabilities, according to the different demands of end customers, establish different business scenarios, deep integration of products and chips, and develop intelligent driving application solutions that integrate driving, parking and parking.
Moshi Intelligent Yu Zhenghua also attended the summit, Moshi Intelligent is the "embedded artificial intelligence autonomous driving enabler", the current stage for passenger cars and commercial vehicles of mass production vehicles to provide camera-based perception technology solutions, to help vehicles to achieve lane departure warning, forward collision warning, emergency braking, automatic parking and other advanced driving assistance functions, and continue to upgrade to L4 automatic driving.
Picture note: Magic Vision Intelligence Yu Zhenghua (middle) and Aiji Wei Chen Yunrong and author Dai Hui
The history of artificial intelligence is also interesting. In 1994, I was doing a graduation project in the image room of UTokyo University, and a teacher was using the pattern recognition method to do the recognition of printed numbers and chromosomes. Around 2010, TI's camera chip integrated a license plate number recognition algorithm based on the DSP. Later, HiSilicon's camera chip integrated the license plate number recognition algorithm developed by Professor Zhao Yong's team at Peking University, using a cheaper ASIC method, based on the basic principles of pattern recognition and feature analysis. Many companies look like hardware or equipment, but in fact, the core is software algorithms, such as my consultant Octavia ideal AOI, which serves Tesla, BMW, Toyota Bosch and BYD and other domestic manufacturers.
With the great improvement of computing power, deep learning (deep learning) has made breakthroughs in North America, 2015 GPU Conference, 2016 CES, technology veteran Dai Hui witnessed this great technological revolution. Unfortunately, I didn't buy stocks, and Zhu Jun, who attended the meeting with me, both sighed.
Photo note: In 2015, author Dai Hui visited NVIDIA
Israeli MOBILE-EYE was the first company to start in the field of ADAS, and was later acquired by intel privatization. The strategy for success is to emphasize software and chip bundling, and to analyze driving data, which has entered most mainstream car factories. After Tesla entered the market with MOBILE-EYE, it began to develop itself and became a trend for many car manufacturers. Recently, MOBILE-EYE has also said that it will decouple, open the road to opening up, and will re-list independently.
Artificial intelligence is also surging Chinese mainland, including security face recognition, beauty mobile phones, car assisted driving and other fields. A large amount of data involving privacy is generated in the local area, and deep learning in the local area has become inevitable.
Chinese mainland ADAS software community has risen to serve the Chinese mainland homegrown automotive industry. Engaged in the ADAS business are also known in the industry as MAXIEYE and MINIYE, which are nicknamed big and small eyes, as well as a number of excellent companies such as Zhongke Huiyan.
On the basis of computing power and deep learning algorithms, data and applications have become the core of the development of artificial intelligence, which is not lacking in China.
VII. Conclusion
At this summit, SAIC, BAIC, local chips and ecological supply chain partners exchanged behind closed doors, and had this unforgettable group photo.
Picture note: Group photo of some participants in the automotive supply chain exchange conference
Let us embrace this era of intelligence with open arms, and there is a song "Boys Should Be Self-Reliant" that sings:
Proud to face the waves
Blood is like that red sun
Gall is like iron, bone is like steel
Broad-minded and long-sighted
I strive to be a good man
Be a good man And be self-reliant every day
The hot-blooded man is brighter than the sun
Let Haitian gather energy for me
Go break new ground
Break through for my ideals
(Proofreader/Fan Rong)