If the car is compared to the human body, the mechanical structure of the car is equivalent to the human skeleton, the power and steering are equivalent to the human limbs, and the electronic and electrical architecture is equivalent to the human nervous system and brain, which is the key to the information interaction and complex operation of the car.
The electronic and electrical architecture covers the software and hardware, sensors, communication network and electrical distribution system of the on-board calculation and control system. It combines all subsystems in order through specific logic and norms to form an organic whole to realize complex functions.
In the era of functional cars, once the car leaves the factory, the user experience is basically solidified; In the era of smart cars, cars are often new and there are thousands of people. The evolution of electronic and electrical architecture to centralization is the premise of this change.
From distributed to domain control to centralized, with the development of chip and communication technology, the electronic and electrical architecture is undergoing tremendous changes.
1.1 The distributed electronic and electrical architecture is overwhelmed. At the beginning of the car’s birth, it was a purely mechanical product. There was no battery on the car, and the equipment on the car did not need electricity. In 1927, Bosch developed lead-acid batteries, and from then on, the electronic equipment on the car had a reliable power source. With the development of large-scale integrated circuits, automotive electronics have developed rapidly. Engine timing ignition control system, electronically controlled fuel injection system, automatic transmission control system, traction control system, electronically controlled suspension system, electronically controlled seat, electronically controlled window, instrument, electronically controlled air conditioner and automotive electronic stability control system have gradually become indispensable components of automobiles. Automotive electronic control technology has gradually developed and expanded, providing consumers with higher performance, more comfortable and safer travel tools. Under the early distributed electronic and electrical architecture, each ECU is usually only responsible for controlling a single functional unit, which is independent of each other and controls the engine, brakes, doors and other components, such as engine controller (ECM), transmission system controller (TCM), brake controller (BCM) and battery management system (BMS). The ecus are connected by CAN(Controller Area Network) bus or LIN(Local Interconnect Network) bus, and exchange information through the communication protocol defined by the manufacturer in advance. With the increasing application of vehicle electronic and electrical products,The number of ecus has rapidly increased from dozens to more than 100. The more ecus there are, the longer the wire harness length of the corresponding bus will be, and the weight of the wire harness will also increase accordingly (the bus length of Audi Q7 and Porsche Cayenne listed in 2007 is over 6km, and the total weight is over 70kg, which is the part of the whole vehicle that is second only to the engine), which leads to the increase of the cost of the whole vehicle and the low level of automation of automobile assembly. Distributed computing leads to the information island in the car, the waste of computing power, the deep coupling between software and hardware, and the OEM relies heavily on suppliers. In the traditional automobile supply chain, different ECU comes from different suppliers, and different hardware has different embedded software and underlying code. The whole vehicle software is actually a mixture of many independent and incompatible software, which leads to the lack of compatibility and expansibility of the whole system. Any function change of a car factory needs to negotiate with many different suppliers about the coordinated development of software and hardware. Every time a new function is added, an ECU and communication system need to be added, which takes a long time and the process is cumbersome. Moreover, because each ECU is bound with a specific function, it is impossible to realize complex functions spanning multiple ECU/ sensors, and it is also impossible to keep the continuous update of automobile software through OTA(Over-the-Air). Distributed electronic and electrical architecture leads to communication bandwidth bottleneck. The function of intelligent networked vehicles is becoming more and more complex, and the number of vehicle sensors is increasing, which leads to the improvement of real-time requirements for data transmission and processing, and the amount of network communication data in automobiles is increasing exponentially.Traditional FlexRay, LIN and CAN low-speed buses can’t provide high-bandwidth communication capability, and can’t meet the real-time requirements of data transmission and processing. We use a concrete example to illustrate the disadvantages of the distributed electronic and electrical architecture: suppose a car factory needs to modify the function of a wiper assembly, because each car needs to define, calibrate and verify the wiper assembly at a given node in the development process, and subsequent modification is equivalent to secondary development, so car companies need to re-sign contracts with wiper suppliers and re-calibrate and verify at all levels. Obviously, such a hardware-oriented engineering system and process can not support the rapid iterative evolution of products in the future when vehicles become more and more complex. The solution is to standardize hardware. The wiper assembly is a mechanical part driven by a motor, and the sensor required by the wiper can call the camera or other sensors mounted on the vehicle. Once the transparency of the windshield is reduced, the vehicle can automatically start the appropriate working mode through software control, thus realizing the purpose of defining the wiper function by software. When various assemblies and modules are standardized, a higher level of intelligence can be realized through the software in the central controller, just like multiple apps running on mobile phones, which can not only greatly shorten the product development cycle, but also widely adopt standardized parts and components, which will help enterprises control costs and quality. For example, a parts company develops and produces a standardized wiper, and then sells it to various vehicle companies, and its price will be very cheap; At the same time,The calibration and verification of standardized hardware can be simplified appropriately, thus further saving development time and cost. 1.2 The automotive electronic and electrical architecture is moving towards central computing. The automotive distributed electronic and electrical architecture can no longer adapt to the further evolution of automotive intelligence. High integration is the solution. Based on a small number of high-performance processors, the "brain" of the car is built, and through a new set of electronic and electrical architecture, the "neural network" and "blood vessel" that can quickly transmit information are formed to control and drive all electronic components and sensors. A small number of high-performance computing units replace a large number of distributed MCU (micro-control units) in the past, and a plurality of scattered small sensors are integrated into a single sensor with stronger functions, and the automobile and functions are gradually integrated and concentrated. The burden reduction of ECU means that dozens of ecus originally carried by the whole vehicle are stripped of software and hardware one by one. Then the functions are mainly migrated to the domain controller through software (domain controller refers to the general name of the whole system composed of domain master hardware, operating system, algorithm and application software), such as automatic driving, entertainment, gateway, etc. On the basis of the domain controller architecture, the domains with different functions are further integrated, and the cross-domain integration stage is reached, and then the central computing+location domain stage is further reached. Huawei judges that by 2030, the electronic and electrical architecture will evolve into the computing and communication architecture of central computing platform+regional access+large bandwidth vehicle communication.The upgrading of automotive electronic and electrical architecture is mainly reflected in hardware architecture, software architecture and communication architecture: hardware architecture develops from distributed to domain control/centralized, software architecture develops from high coupling of software and hardware to layered decoupling, and communication architecture develops from LIN/CAN bus to Ethernet. The roadmap of electronic and electrical architecture given by Bosch is divided into six stages, which has become an industry consensus: distributed stage (including modularization and integration)-domain centralization (including centralization and domain integration) and central centralization (including on-board computers and vehicle cloud computing). Modularization stage. 1) An ECU is responsible for specific functions, such as a controller for the lights on the car, a controller for the doors and a controller for the keyless system. With the increase of automobile functions, this architecture becomes increasingly complex and unsustainable. 2) In the integration stage, a single ECU is responsible for multiple functions, and the number of ECU is reduced compared with the previous stage. In these two stages, the automotive electronic and electrical architecture is still in the distributed stage, and the functional integration of ECU is low. Functional domain control stage. Functional domains are domain controllers divided according to functions, and the most common ones are five functional domains (power domain, chassis domain, body domain, cockpit domain and autopilot domain) divided by Bosch. The domain controllers are connected with CANFD(CAN with Flexible Data-Rate) through Ethernet, and the demand for computing power in cockpit domain and autopilot domain is increasing gradually due to the large amount of data to be processed.Powertrain domain, chassis domain and body domain mainly involve control command calculation and communication resources, and the calculation requirements are low. Cross-domain integration stage. On the basis of functional domains, in order to further reduce costs and enhance collaboration, cross-domain integration has emerged, that is, multiple domains are integrated and controlled by cross-domain control units. For example, power domain, chassis domain and body domain are merged into vehicle control domain, so that five functional domains (autopilot domain, power domain, chassis domain, cockpit domain and body domain) are transformed into three functional domains (autopilot domain, intelligent cockpit domain and vehicle control domain). Central computing+location domain stage. With the deep integration of functional domains, functional domains are gradually upgraded to a more general computing platform, and they move from functional domains to location domains (such as middle domain, left domain and right domain). The Zonal Control Unit (ZCU) is a local sensing, data processing, control and execution unit in the vehicle computing system. It is responsible for connecting sensors, actuators and ECU in a certain area of the vehicle, and is responsible for the preliminary calculation and processing of sensor data in this location area, and is also responsible for the network protocol conversion in this area. The wiring harness can be arranged nearby in the position domain, which reduces the cost and communication interface, and makes it easier to realize the automatic assembly of the wiring harness, thus improving the efficiency. Sensors, actuators, etc. are connected to nearby regional controllers, which can better realize hardware expansion and make the structure management of regional controllers easier. Regional access and central computing ensure the stability of the vehicle architecture and the expansion of functions.The newly added external components can be accessed based on the regional gateway, the pluggable design of hardware supports the continuous improvement of computing power, and sufficient computing power supports the iterative upgrade of application software on the central computing platform. In a research aimed at a vehicle manufacturer, Amber found that nine ecus can be integrated by using the regional controller, and hundreds of individual wires can be saved, thus reducing the weight of the vehicle by 8.5 kg. Reducing weight helps to save energy and extend the driving range of electric vehicles. In addition, because the regional controller divides the basic electrical structure of the vehicle into more manageable components, it is easier to realize automatic wiring harness assembly. Automobile cloud computing stage. Transfer some functions of the car to the cloud, and further simplify the interior structure. Various sensors and actuators of the car can be defined and controlled by software, and the parts of the car will gradually become standard parts, thus completely realizing the function of software-defined car. The evolution of automotive electronic and electrical architecture provides strong support for software and hardware decoupling. The highly centralized electronic and electrical architecture brings centralized calculation, software and hardware decoupling, platform standardization and function customization. 1) The computing power tends to be centralized, and many ecus are concentrated on several powerful computing platforms, which provides the computing power foundation for software operation; 2) The underlying software and code began to get through, the software ecosystem with the operating system as the core began to be established, the software could realize continuous iteration, and the development of OTA was accelerated; 3) Domain controller and time-sensitive Ethernet can realize high-speed data processing and transmission, which creates conditions for the development of software applications. Comparison of the progress of electronic and electrical architecture in various main engine plants
In the future, the core technology of automotive products is electronic and electrical architecture, which is gradually developing from decentralized and embedded to centralized and integrated. The ultimate ideal state should be to form an automotive one brain to manage various functions in a unified way.
The electronic and electrical architecture is similar to the "central government", which can manage all kinds of functions of the car as a whole and avoid "different governors and decrees". At the beginning, the "central government" may manage less, and the "local governors" still retain some control rights, but after that, the "central government" will certainly manage more and more. In the end, the local administrative agencies will only receive the instructions from the "central government" and implement them efficiently to ensure the overall performance of the vehicle. In the past, because the controllers on the car were independent of each other and the software was embedded, the final hardware integration of the whole car was enough. In the future, with the burden reduction of ECU, the original highly decentralized functions will be integrated into the domain controller, and the main engine factory must master the central control system by itself, otherwise it will lose the control right of automobile products. It is a new compulsory course for traditional car companies to gradually integrate and unify the original highly decentralized control functions, so the car companies’ mastery of electronic and electrical architecture is step by step and gradual. Tesla Model3 has initiated a great change in the electronic and electrical architecture, with the emergence of the prototype of central computing+location domain, shortening the vehicle wiring harness by 50%. The future goal is to reduce the vehicle wiring harness to 100 meters. In terms of electronic architecture, Tesla is ahead of traditional car companies for more than 6 years. Except Tesla, most of the electronic and electrical architectures of automobile enterprises are still in the early stage of functional domain controller, that is, some functions are concentrated in functional domain controller, but there are still more distributed modules, that is, the transition scheme of "distributed ECU+ domain controller", to avoid additional risks and costs caused by too much change.The next generation of cross-domain integrated electronic and electrical architecture planned by most enterprises will be mass-produced in 2022, so as to realize the high concentration of software in domain controllers and gradually reduce distributed ECU. By 2025, some car companies will have the electronic and electrical architecture of central computing+regional controller, so as to realize further integration of software and hardware, and the software ownership will gradually be returned to the OEM. The evolution towards the architecture of "central computing+regional control" may take as long as 5-10 years. 2.1 Audi A8 The Audi A8 launched in 2018 took the lead in realizing the integrated control of the assisted driving function, replacing the distributed assisted driving system with ECU separated from each other. In addition to the integration of autonomous driving domain, the other four domains of chassis+safety, power, body and entertainment still adopt distributed architecture. Its autopilot controller consists of four chips, and Mobileyeye EQ 3 is responsible for visual perception calculation, such as traffic signal recognition, pedestrian monitoring, collision alarm, lane line recognition and light detection. Nvidia K1 is responsible for image fusion calculation, such as driver monitoring and image processing of 360 panoramic camera. Intel Cyclone V is responsible for target fusion, map fusion, parking assistance and pre-brake lights. Aurix TC297 of Infineon is responsible for communication processing. The software development of this autopilot controller was completed by TTTech, an Austrian software company, and Delphi provided hardware integration.2.2 Tesla Model3 opens a comprehensive transformation of the electronic and electrical architecture. Tesla is a comprehensive reformer of the automotive electronic and electrical architecture. In 2012, Model S had obvious functional domain division, including power domain, chassis domain and body domain. The ADAS module spanned the power and chassis domain. Because the traditional domain architecture cannot meet the development of autonomous driving technology and the needs of software-defined cars, in order to decouple software and hardware, To carry a more powerful main control chip, the electronic and electrical architecture must be reformed first. Therefore, in 2017, Model3 launched by Tesla broke through the framework of functional domain, and realized the framework of central computing+regional controller. By building an exotic fusion architecture+independent software platform, not only the software defined the car, but also the cost of the whole vehicle was effectively reduced and the efficiency was improved: 1)Model 3 had three controllers, which effectively reduced the material cost; 2) Hardware integration is software, which provides the basis for the control and maintenance of vehicle depth; 3) Independent software platform supports extended reuse through modularization. Tesla Model3 basically realized the prototype of centralized architecture, but Model3 is still a long way from the real centralized architecture: the communication architecture is mainly based on CAN bus, and the central computing module only integrates audio-visual entertainment MCU, autonomous driving FSD and in-vehicle networking module on one board, and each module runs its own operating system independently. But anyway,Model3 has implemented the conceptual framework of electronic and electrical architecture of central computing+regional control, leading traditional car companies for about 6 years. The essence behind the evolution of the electronic and electrical architecture of Tesla’s third-generation cars is the process of constantly taking back the vehicle functions from suppliers and developing them independently. Model3′ s autopilot module, entertainment control module, other area controllers and thermal management are all independently designed and developed, which realizes the autonomy of the main modules of the whole vehicle and does not depend on Tier1. Even if there is no independent module, Tesla has carried out joint development with suppliers. For example, Tesla added its own software to ibooster provided by Bosch, and shortened the braking distance through software update. Through the evolution of three MODEL S, Tesla’s new electronic and electrical architecture has not only greatly reduced the number of ECU and the wiring harness (the wiring harness of Model S is 3,000 meters, and the wiring harness of Model 3 is reduced by more than half), but also broken the old parts supply system of the automobile industry (that is, the software and hardware are deeply coupled and packaged and sold to the OEM, and the OEM’s bargaining power is poor, and the subsequent function adjustment is difficult), thus truly realizing the software-defined automobile. Tesla’s OTA can change the braking distance, turn on the seat heating, and provide personalized user experience. Due to the breakthrough of the functional domain, Tesla’s domain controller spans the body, cockpit, chassis and power domain, which makes the functional iteration of the vehicle more flexible. Users can experience that the car is often used and always new. In sharp contrast,OTA in most traditional car factories is limited to in-vehicle infotainment and other functions. In order to give full play to the role of software, Tesla has realized the self-research and self-manufacture of the core intelligent hardware, namely, the autopilot master control chip (Tesla believes that the special design of the chip makes the software on it run more efficiently), which means that the upgrade speed and function deployment of subsequent Tesla vehicles no longer depend on external SOC chip suppliers, and the soul of the vehicle is really in their own hands. The four controllers of the Model 3 vehicle include four domain controllers: Central Computing Module (CCM), Left Body Control Module (BCM LH), Right Body Control Module (BCM RH) and Front Body Control Module (BCM FH). The left body control module is responsible for the convenience control of the left body, steering, braking, power assistance, etc. The right body control module is responsible for the right body convenience control, chassis safety system, power system, thermal management, etc. The central computing module includes an automatic driving module, an infotainment module, and communication connections inside and outside the vehicle, sharing a set of liquid cooling system. The automatic driving and entertainment control module takes over the sensors related to assisted driving-camera and millimeter-wave radar, and puts intelligent driving and infotainment with high computing power requirements together, which is convenient for the continuous upgrade of intelligent hardware. In 2019, Tesla introduced the self-developed FSD chip to replace the NVIDIA Drive PX2 chipset, which improved the AI computing performance by 21 times. With Tesla’s self-developed computing hardware,Tesla has greatly improved its lead over its competitors. The operating system is customized based on open source Linux, and the middleware, software and hardware are self-controllable, which speeds up the iterative update of vehicle functions and reduces the development cost of the whole vehicle. 2.3 Volkswagen ID series electronic and electrical architecture Volkswagen has upgraded from the distributed electronic and electrical architecture of MQB platform models to the electronic and electrical architecture of three functional domains adopted on MEB platform ID series models. According to the plan, the electronic and electrical architecture of ID series based on Volkswagen MEB platform is version E31.1, and version E31.2 will be installed on PPE platform in 2023, and it will not evolve to version E32.0 until 2025. Volkswagen’s E3 architecture is mainly composed of vehicle control domain (ICAS1), intelligent driving domain (ICAS2) and intelligent cockpit domain (ICAS3). Among them, the intelligent driving domain ICAS2 has not been developed yet, and the mass production vehicles are still equipped with distributed architecture scheme. Although the electronic and electrical architecture of Volkswagen ID series has three functional domains, it still retains more distributed modules. Volkswagen ID4 has 52 ecus, twice as many as Tesla Model Y ECU. The assisted driving function of domestic ID4 is realized by Mobileye monocular camera+front long-range radar+two rear-angle radars. As a cheap electric vehicle, we have no choice to go to PK with Tesla and China’s new forces in the automatic driving domain controller for the time being.Volkswagen ID series vehicles will deliver 70,000 units in 2021, which is lower than the previous planning. China, as the most important single market for Volkswagen, is also catching up with intelligence. In 2022, CARIAD, a software company of Volkswagen, set up a subsidiary in China. According to the CEO of its China subsidiary, the company’s core business is software research and development for MEB platform, and OTA function will be launched in the second half of 2022. Secondly, China localization and digital products, including advanced driver assistance systems, will be made for high-end platforms (PPE’s first car in China will be put into production in 2024), and its intelligent networking system will also be connected with China. The third is to do software research and development around SSP platform after 2025. According to the planning of Volkswagen 2030 NEW auto, the proportion of self-developed software will rise to 60%. The advantages of maintaining independent software research and development are agility (including development and maintenance) and product differentiation. Localization is also a necessary and key link for foreign capital to enhance intelligence in China, with the ultimate goal of creating competitive products that attract China users. Let’s take a look at the comparison of the electronic and electrical architectures of three electric vehicles that came out at the same time. Although the Volkswagen ID series also claims to replace the past 70+ distributed ECU with three domain controllers, it actually still maintains a large number of ECU’s. Before ID3, it was delayed due to a large area of software bugs.This also reflects that even if the traditional car factory chooses to carry out major changes in the electronic and electrical architecture, if its own talent structure and software strength are not enough, it will still rely heavily on external suppliers, resulting in excessive steps and additional risks. Therefore, most OEMs choose to take a gradual route and gradually return to the software dominance with the improvement of their own software strength. In 2021, Munro & Associates Engineering Company compared the differences between Tesla Model Y, Ford Mach-E and Volkswagen ID.4 electrical architectures. It involves the number of ECU, CAN bus, Ethernet, LIN bus, LVDS (low-voltage differential signaling) channel, audio, fuse and relay. Tesla Model Y is obviously more integrated, and its ECU number is half that of ID4. Ford and Volkswagen still keep more ready-made distributed ECU, and Tesla’s LIN (Local Interconnection Network) number is only half that of Volkswagen ID4 and Ford Mach-E.. The number of CAN (Controller Area Network) buses in Tesla is higher. Due to the increase in the number of cameras, Tesla’s low-voltage differential signal (LVDS) usage is more than three times that of Ford and Volkswagen, and Volkswagen’s Ethernet usage is more.Tesla does not use any fuse box relay in the low-voltage electrical part of the vehicle from Model 3. 2.4 Xpeng Motors G9 electronic and electrical architecture has a leading position. Among the top three new forces, Xpeng Motors has taken the lead in electronic and electrical architecture. With the iteration of the X-EEA3.0 electronic and electrical architecture from G3, P7 and P5 to G9, it has entered the centralized electronic and electrical architecture. With the leading generation architecture, higher computing power SOC chip and higher computing power utilization rate, Tucki G9 may become the first production car supporting XPILOT 4.0 intelligent assisted driving system. Tucki P7 is equipped with Tucki’s second generation electronic and electrical architecture, which has hybrid characteristics: 1) Hierarchical domain control. Functional domain controllers (intelligent driving domain controller, body domain controller, power domain controller and other modules) coexist with the central domain controller; 2) Cross-domain integration-domain controller covers multiple functions and retains local traditional ECU；; 3) Hybrid design-traditional signal interaction and service interaction become a coexistence design. Therefore, CAN bus and Ethernet bus coexist, and big data/real-time interaction can be guaranteed. There are few Ethernet nodes and low requirements for gateways. Tucki’s second-generation electronic and electrical architecture reduces the number of traditional ECU by about 60%, realizes high integration of hardware resources, migrates most car body functions to domain controllers, and the central processor can realize most functions of supporting instruments, infotainment systems and intelligent car body related controls.At the same time, it integrates the central gateway, is compatible with the V2X protocol, supports the communication between vehicles and the local area network, supports the interconnection between vehicles and the cloud, and supports the connection function between vehicles and remote digital terminals. Xpeng Motors’s intelligent driving domain controller integrates high-speed NGP, urban GNP and parking functions. Tucki uses lidar vision fusion scheme for assisted driving, which is different from Tesla’s pure vision scheme, which leads to different hardware architectures and different requirements for communication bandwidth and computing power. Xpeng Motors called its X-EEA3.0 electronic and electrical architecture "the secret to make smart cars never fall behind in the future". According to the information disclosed by the company on the electronic and electrical architecture of G9, G9 has great potential for upgrading and optimization in the future. In the hardware architecture of X-EEA 3.0, the hardware architecture of Central Supercomputer (C-DCU)+ Zone Control (Z-DCU) is adopted. The Central Supercomputer includes three domain controllers, namely, vehicle control, intelligent driving and cockpit, and the zone controllers are left and right domain controllers, which divide more control parts, and according to the principle of nearby configuration, the zones take over the corresponding functions, greatly reducing the wiring harness. Thanks to Xiaopeng Automobile’s full-stack self-research capability, the new architecture has achieved deep integration of hardware and software, which not only realizes decoupling of software and hardware, but also realizes layered decoupling of software, which can make the system software platform, basic software platform and intelligent application platform iterate in layers, and separate the underlying software and basic software of the vehicle from the application software related to intelligence, technology and performance. When developing new functions,It only needs to research and iterate the top application software, which shortens the research and development cycle and technical barriers, and users can also enjoy the rapid iteration of the car. System software platform: partially customized based on outsourced codes, frozen with the freezing of the basic software platform of the whole vehicle, which can be reused for different models; Basic software platform: a number of basic functional software of the whole vehicle form standard service interfaces and freeze before mass production of the vehicle, which can be reused for different models; Intelligent application platform: such as automatic driving, intelligent voice control, intelligent scene and other functions, which can realize rapid development and iteration. In the data architecture of X-EEA 3.0, the domain controller sets the memory partition, and the upgrade operation does not interfere with each other, so the upgrade can be completed by car side in 30 minutes. In terms of communication architecture, X-EEA3.0 has realized the communication architecture with Gigabit Ethernet as the backbone for the first time in China, and supports multiple communication protocols, which makes vehicles faster in data transmission. As can be seen from the new generation of electronic and electrical architecture carried by G9, Tucki started early in the construction of backbone network and SOA-oriented direction. In terms of X-EEA 3.0 power architecture, it can realize scene-based precise power distribution, and can distribute power according to different car scenes such as driving and the third space. For example, when waiting for people on the roadside, it can only supply power to functions such as air conditioning, seat adjustment, music, etc., and other parts are powered off, so that energy consumption can be saved and cruising range can be improved. Vehicles regularly self-diagnose, actively find problems, guide maintenance, and empower after-sales by scientific and technological means.2.5 Development Roadmap of Electronic and Electrical Architecture of Great Wall Motor The third-generation electronic and electrical architecture developed by Great Wall Motor in 2020 includes four functional domain controllers-body control, power chassis, intelligent cockpit and intelligent driving. The application software has been independently developed, which has been mass-produced and applied to all models of Great Wall Motor, and the material cost of the models has been optimized. For example, the new Haval H6 has optimized the 300-meter wiring harness, with a total length of 1.6 kilometers, which is close to Tesla Model 3, with a weight loss of over 2 kilograms. Since GEEP3.0, Great Wall Motor has realized the independent development capability of all application-layer software, and the upper-layer application software of four domain controllers and even some bottom-layer and bottom-layer integrated software have also been independently developed by Great Wall Motor. The fourth generation electronic and electrical architecture to be launched in 2022 will further centralize vehicle control software, and realize efficient integrated management, high safety and reliability, and faster demand response. The fourth generation architecture has three computing platforms: central computing, intelligent cockpit and advanced autopilot, plus three regional controllers (left, right and front). The fourth generation architecture will be the first to be carried on the brand-new electric and hybrid platform of Great Wall Motor, and will be extended to all models one after another. The central computing unit of the fourth generation electronic and electrical architecture integrates the functions of car body, gateway, air conditioner, power/chassis control and ADAS across domains, and its main control chip has a computing power of up to 30KDMIPS, which can effectively guarantee the control and response of the system. GEEP 4.0 architecture has a mature vision processing chip solution.18-way CAN FD, 4-way LIN, 11-way car Ethernet, 64GB storage and 1GB memory, etc., to meet the needs of computing power and communication brought by future functional integration. The three regional controllers are standardized control units, which are responsible for integrating the peripheral MCU. At present, most of the software algorithms of the three regional controllers have been moved to the central computing unit and developed by the Great Wall software team. The architecture introduces SOA design methods and concepts, creates a layered software infrastructure platform, and provides modular standard service interfaces. The advantages of this architecture are that it can provide modular disassembly and assembly, decouple software and hardware platforms, improve software reusability, and enable the car to realize iterative upgrade of functions throughout its life cycle. Users can dynamically subscribe to upgrade vehicle service functions according to their needs and preferences without waiting for software upgrade batches. At the same time, SOA can also flexibly deploy intelligent scenes, standardized interfaces can realize open services, build a great wall motor creative ecology, and co-developers can provide users with full-scenario smart travel services. GEEP 4.0 supports firmware air upgrade, software air upgrade and remote diagnosis; At the same time, it supports all ECU OTA functions of the whole vehicle, including power chassis system, audio-visual entertainment system, body system and intelligent driving system. The cloud diagnosis method based on the brand-new architecture brings convenience to after-sales service. Based on the deployment of vehicle-side and cloud functions, the vehicle fault information can be diagnosed remotely, and the vehicle can be repaired remotely. While ensure that timeliness of diagnosis and maintenance,Through the diagnosis knowledge base, we can intelligently identify, analyze and match the optimal maintenance scheme, effectively solve the shortcomings of insufficient staff and limited technology in 4S stores, and truly solve problems for users quickly. Great Wall Motor’s fifth-generation electronic and electrical architecture research and development started simultaneously with the fourth generation. The fifth-generation architecture highly concentrates the whole vehicle software in one brain, and it is planned to be available in 2024. It will realize 100% SOA and complete the construction of vehicle standardization software platform. At present, the cockpit chip and the intelligent driving chip of the central computing module used by Tesla are separated, which is not the one brain scheme. From the current trend of the global head intelligent chip manufacturers, it is the general trend that the intelligent driving chip and the cockpit chip are integrated, but the one brain scheme requires high software capabilities of the main engine factory. Great Wall Motor’s electronic and electrical architecture has a fast iteration speed, which will provide a "foundation" for the self-developed intelligent core technology. The rapid iteration of electronic and electrical architecture is also strongly related to the company’s goal of maintaining a leading position in intelligence. In terms of intelligence, the typical winning weapons of the Great Wall are: 1) the self-driving full-stack self-research technology of Zhixing. 2) Steering-by-wire technology put into commercial application in 2023. Full-stack self-research of autonomous driving solution: Great Wall Motor’s Millimeter Smart will realize the urban pilot-assisted driving function in 2022, or compete with Xpeng Motors for the landing rhythm of urban pilot function. On the hardware side,HPilot3.0 has a strong computing power of 360TOPS, and the whole vehicle is equipped with 12 cameras and 2 laser radars, 5 millimeter-wave radars and 12 ultrasonic radars. One of the reasons why Mimo Zhixing’s urban navigation function took the lead in landing is that it adopts the scheme of re-perception, not the scheme of re-map, which is not limited by the high-precision map of the city. In June, 2022, Mimo Zhixing City Pilot Plan SOP can be effectively deployed in more than 100 cities across the country, which has great advantages in geographical scope. Mimo Zhixing has a large overall deployment range, many models and a large number, and can maintain high-speed continuous iteration based on more data. In 2022, it will undertake the task of developing high-level assisted driving for Great Wall Motor’s 34 hospitality listed models, accounting for nearly 80% of the models to be listed in the whole year, 30% of which are standard, and the rest are equipped with high equipment. Automatic driving implementation: intelligent upgrading of automobiles and centralization of electronic and electrical architecture, and at the same time, it is necessary to upgrade the traditional automobile chassis by wire to adapt to the development. The chassis control system is strongly related to the implementation of automatic driving. The chassis by wire mainly includes steering by wire, braking by wire, gear shifting by wire, throttle by wire and suspension by wire, among which steering by wire and braking by wire are the core products for automatic driving. At present, the main brake by wire manufacturers in the world are traditional Tier1 such as Bosch, Continental and ZF, and the entry threshold is very high. In 2021, China Great Wall Motor first released the intelligent drive-by-wire chassis.From the core hardware such as electro-mechanical brake by wire, steering gear, motor, simulator and controller to the whole software system, Great Wall Motor independently designed and completed it. This is the first steer-by-wire technology in China that supports L4+ autopilot, and it will be put into commercial application in 2023. 2.6 SAIC Zero-beam Electronic and Electrical Architecture Zu Sijie, the chief engineer of SAIC, believes that the core technology of automotive products is electronic and electrical architecture, and it must be mastered by vehicle manufacturers. As the hub of the automobile, the electronic and electrical architecture will define many related standards that are completely different from those before, because the automobile was a closed system in the past, and the automobile will be an open system in the future. After the popularization of self-driving cars, car companies should bear the responsibility of traffic safety accidents, and the safety technology can only be mastered by themselves. From this point of view, car companies should also firmly grasp the electronic and electrical architecture and central control system in their own hands, including the on-board operating system, basic application and service software architecture above the electronic and electrical architecture, which should be fully understood and integrated. From the perspective of vehicle product control, Zu Sijie believes that the controllers on the original vehicle products are independent of each other and embedded, so it won’t be a big problem for vehicle companies to hand over some of them to suppliers. In the future, the control systems on vehicle products will be unified, and vehicle companies must master the central control system by themselves, otherwise they will lose control over the vehicle products. It is a correct and difficult way for car companies to gradually integrate and unify the originally highly decentralized control functions.SAIC is equipped with full-stack version 1.0 electronic and electrical architecture in its high-end pure electric smart car brands Zhiji and Feifan. The full-stack version 1.0 electronic and electrical architecture has three domain controllers, namely, central computing (vehicle control and data fusion), intelligent driving and intelligent cockpit, while retaining more distributed modules. In July 2021, the independent research and development of "Zero Beam Galaxy Full Stack 3.0 Technical Solution" was launched, which further centralized and supported automatic driving above L4 level. It is planned to be carried by Zhizhi and Feifan under SAIC in 2024. The electronic and electrical architecture of Zero Beam Galaxy Full Stack 3.0 uses two high-performance computing units, namely HPC1 and HPC2, to realize the functions of intelligent driving, intelligent cockpit, intelligent computing and intelligent driving backup, and four regional controllers are added to realize the related functions in different regions, so as to fully support the intelligent driving technology above L4. The bottom narrow operating system (OS) is upgraded from heterogeneous to isomorphic; The backbone communication bandwidth is expanded to gigabit or even 10 trillion; The smart car data factory fully realizes the digital twin mirror image, continuously consolidates the network security protection system of smart cars in cloud, tube and end, accelerates the self-learning, self-growth and self-evolution of smart cars, and makes the cars truly become carriers and portals of directly connected users, mobile AIoT platforms and digital experience spaces. 2.7 Guangzhou Automobile Spirit Electronic and Electrical Architecture The electronic and electrical architecture of Guangzhou Automobile Spirit is planned to be carried on the new model of Guangzhou Automobile Ai Safety in 2023, which is composed of automobile digital mirror cloud,Central computer, intelligent driving computer, infotainment computer and four regional controllers, which integrate Gigabit Ethernet, 5G, information security, functional security and other technologies. Compared with the previous generation electronic and electrical architecture of GAC, the computing power of the new architecture is increased by 50 times, the data transmission rate is increased by 10 times, the harness loop is reduced by about 40%, and the number of controllers is reduced by about 20. The hardware architecture has three functional domain controllers+four regional controllers, which is similar to the fourth generation electronic and electrical architecture of Great Wall Motor. The central computing unit (body control+new energy control) is equipped with NXP S32G399 high-performance gateway computing chip; The cockpit area is equipped with Qualcomm 8155/8295 chips; Intelligent driving domain is equipped with Huawei Shengteng 610 high-performance chip with a computing power of 400TOPS. Four regional controllers distributed in the front, back, left and right of the car body are mainly responsible for power supply and executing the instructions of the central control unit, and the central computing unit is connected with the four regional controllers by Ethernet. In terms of software structure, "Spirit" architecture adopts SOA software architecture to replace the traditional software architecture, so as to realize service, atomization and standardization of components, and new scenarios can be realized by adding application modules. A good electronic and electrical architecture can save costs, including manufacturing costs and car costs, and the production end can save materials, simplify assembly and improve development and manufacturing efficiency. Under the condition of similar surface functions, consumers may consume less energy when using cars with higher electronic architecture integration.The second is to quickly provide a variety of functions. The OEM can develop various functions for different scenarios, such as Tesla’s seat heating and holiday mode, and the function update should be controlled by the OEM, without having to organize a complex supply chain to change a function like the previous function car. Without the upgrade of the underlying architecture, no matter how many intelligent functions there are on the surface, it can’t be regarded as a real smart car. For example, the distributed electronic and electrical architecture can also realize the functions of automatic parking and L2 intelligent driving. However, due to the limitation of the architecture, the sensor can not be connected to an intelligent driving domain controller, and only two independent control units-parking controller and driving controller can be carried, and the computing power and sensing hardware can not be shared, which leads to a waste of resources and constraints in the subsequent function upgrade. Product definition is the premise of architecture development, and car companies will make choices according to their own brand image, product positioning, target customers and internal resources. For example, car companies may give priority to the integration in the intelligent cockpit, while the auxiliary driving part adopts a low-cost distributed scheme. It is also possible to give priority to highly integrated chassis and body control. There are differences in brand matrix and vehicle structure among different car companies, and the platform commonality and continuity should also be considered in the architecture. Architecture evolution drives multiple changes in OEM.
3.1 Architecture evolution, the ownership of automotive software gradually reverts to the OEM, and the automotive electronic architecture moves towards central computing, and the number of ECU is reduced, which means that the original software and hardware modules are dismantled and then the domain controller is centralized, which is not a simple physical integration. More and more OEMs are gathering more initiative, from application software to middleware, to bottom software, and even to core hardware, hoping to achieve full stack coverage. This process is a process in which OEMs extract the software parts of the original software and hardware suppliers and gather them together. Constrained by the existing supply chain and the weak strength of its own software, this will be a gradual process. Once the electronic and electrical architecture enters the stage of central computing+regional control, the ownership of automobile software will mainly belong to the OEM, which will enjoy the software dividend for a long time and have a stronger say in the industrial chain than in the traditional automobile era, and the OEM will hold the lifeblood of continuous product update in its own hands. Under the distributed architecture, the OEM is equivalent to a hardware integrator. Tier1 packages the upstream Tier2 (embedded software supplier and chip supplier) and provides it to the OEM. In order to improve the product control right, the OEM generally chooses self-developed high-value modules in the era of function cars, which is also a differentiated part that consumers can feel, that is, the powertrain part. In the functional domain part of the second stage, similar functions were merged into domains, and the software was gradually separated from the black box in the past. For the sake of the original supply chain and its own software capabilities, the OEM chose to cooperate directly with the original Tier1/2, in the application software layer.You may choose cooperation mode or self-developed mode. For example, Tucki and Great Wall choose self-developed autopilot algorithms, while other OEMs choose to cooperate with Baidu, momenta, Xiaoma Zhixing and Huawei. At this time, according to their different abilities, OEMs have different degrees of participation in the software and hardware parts of domain controllers, so the services of domain controller suppliers are also different. For OEMs with a deep degree of self-research, domain controller suppliers are equivalent to pure OEM roles, while for OEMs with a shallow degree of self-research, domain controller suppliers are equivalent to all-round "nanny" roles. In the third stage, after stepping out of the functional domain framework and entering the stage of central computing+regional control, most ECU disappears, and all sensors/actuators are dominated by the central computing unit. Most of the software originally belonging to Tier1′ s strategy layer is dominated by the OEM, and the OEM is increasingly involved in the high-value modules in the software. Therefore, the OEM must have a professional software team to integrate self-developed and outsourced software, and the software ownership mainly belongs to automobile manufacturers. Volkswagen plans to increase the self-research ratio of software to 60% by 2030 (focusing on artificial intelligence, big data, confidentiality and security). Although the self-research ratio will be greatly increased, the total scale of outsourced software will also increase, but Volkswagen will define the standards and roadmap of in-vehicle software. CARIAD business planning mainly covers four parts: 1) electronic and electrical architecture E architecture; 2)VW.OS (Volkswagen operating system); 3)VW.AC (Volkswagen Cloud); 4) Key applications.The number of self-developed high-value modules will largely determine the profitability of different OEMs, which is similar to the huge difference in profitability of different consumer electronics brands (the net profit rate of Apple is 20-26% and the net profit rate of Xiaomi is 5-8% in recent three years). The OEM evaluates its own path according to its own business volume, R&D strength, cash flow and historical burden. Parity and luxury, fuel and electricity will all make completely different transformation choices. 1) The first echelon, such as Tesla, realizes self-research and hardware outsourcing in core areas such as chip, operating system, middleware and domain controller system integration. 2) Choose one or two key breakthroughs in core technology. For example, whether the automatic driving perception algorithm chooses self-research. Tucki, Great Wall and Huawei are relatively ahead in terms of the pace of self-research and landing of autonomous driving algorithm. There are also OEMs that choose to develop their own cockpit chips, such as Yikatong, a subsidiary of Geely Automobile. 3) Multi-hand preparation. On the one hand, OEMs set up their own software teams, on the other hand, they actively establish cooperation alliances with technology companies/Internet companies. Before they have mature software development capabilities, the basic software and software and hardware architecture solutions still rely on TIER-1 or emerging software companies. For example, Zero Beam of SAIC currently cooperates with various external enterprises, covering SOC chip companies, algorithm companies and domain controller suppliers (Qualcomm, Horizon, United Electronics, momenta). 4) Car companies only do brand operation, software development is mainly outsourced, and parts are packaged as a whole system to large suppliers or Internet companies.With the evolution of electronic and electrical architecture, the status change of OEM is viewed from three dimensions of safety, data and users: 1) The integrated development of electronic architecture to Che Yun will make smart cars become a more open intelligent contact, and the safety requirements are greatly improved, with OEM as the first responsible body. 2) Choose a self-developed main engine factory for assisted driving, and the assisted driving function will be used more and more. With the gradual evolution of assisted driving, more and more L4 algorithm companies have started to cooperate with OEMs, which also shows that the data generated by effective mileage has become the key point of whether the assisted driving ability can lead in the next stage. 3) OEM will directly link users in real time during the whole life cycle of vehicles, and the stickiness with C-end will be significantly enhanced. User operation, remote diagnosis and service will become the trigger points of OEM’s business based on stock vehicles. The traditional V-shaped R&D model of automobile enterprises (including mechanical hardware testing, supply chain collaboration, styling design, etc.) needs 5-7 years of R&D cycle, which cannot adapt to the rapid iteration of mobile services and the operation of existing users. In the future, the separation of software and hardware research and development will transform the software into a closed-loop development mode with rapid iteration, while the hardware can be embedded in advance and remain unchanged for a long time. The evolution of electronic architecture changes the organizational structure of OEMs: Conway’s law put forward by Marvin Conway in 1967 puts forward that the system/product designed by an organization is the epitome of its internal communication structure. This means that what kind of products/systems an enterprise wants, what kind of organization and organizational culture it needs. In the era of distributed ECU,Car companies only need to do integrated hardware to complete product production and delivery; In the era of functional domain controller, the software was recovered from distributed ECU and moved to functional domain controller, but all departments were still divided according to functions, such as intelligent cockpit, intelligent driving and intelligent vehicle control. The functional domain evolves further. In the era of central computing platform, the hardware is unified and integrated, the software and hardware development are decoupled, the software and layers are decoupled, the adjustment of development team is inevitable, and the chimney-like organizational design divided by functions will be broken. Therefore, in recent years, OEMs have set up digital centers, software centers and other new organizations in order to adapt to this development trend. In 2021, Great Wall Motor will form a version 3.0 organizational structure of "strong background, large, medium and small front desk": strong background is to reserve the best quality and more cutting-edge technology, and keep ahead through a large number of cutting-edge pre-research investments. In addition to technology, the broad background also includes mechanism quality, human resource policy, strategic layout and capital operation. In the battle of the small front desk, Dazhong Taiwan always provides timely supply and support, and takes the small front desk facing users as the core, forming the organizational form of "one car, one brand, one company" and creating a number of operating organizations. At present, car companies are adapting to the evolution trend of electronic and electrical architecture through various organizational innovations. With the evolution of architecture, software centers/technology companies/digital centers of car companies need a lot of R&D investment: Take CARIAD, a popular software company, for example, it plans to develop 60% of its own software, focusing on: 1).Electronic and electrical architecture e architecture; 2)VW.OS (Volkswagen operating system); 3)VW.AC (Volkswagen Cloud); 4) Key applications, with the goal of having 10,000 engineers by 2025. Volkswagen plans to invest 2.5-3 billion euros in CARIAD every year. The evolution of electronic architecture has changed the structure of R&D talents in OEMs: In the future, the proportion of R&D talents in software will increase rapidly. Great Wall Motor announced in mid-2021 that it plans to have 30,000 global R&D personnel by 2023, including 10,000 software development talents. Comparing the R&D expenditure of new forces and traditional car companies, we can clearly see that the newly established car companies are more inclined to software R&D, because they urgently need "full-stack self-research" to form differentiated characteristics in order to get ahead in the brand-new track. Because the average expenditure of R&D personnel is high, the absolute amount of R&D expenditure increases sharply with the increase of the proportion of self-research. For example, Weilai Automobile’s R&D expenditure guideline in 2022 is about 9 billion, which is equivalent to Great Wall Motor’s R&D expenditure in 2021. At present, the development cost of a car body controller in China is 10 million+.According to the prediction of the chief engineer of electronic and electrical architecture of a main engine factory, it will take at least several billion yuan to build a brand-new architecture, develop all the controllers into mass production and deliver them on the market with fewer software bugs. Looking at the investment of software centers of major OEMs, the annual comprehensive cost of a software developer is at least 1 million yuan.Assuming that there are 1000 software teams, the annual staff investment is 1 billion, and if other related inputs such as tool chain are added, the annual expenditure is 2 billion. 3.2 The upgrade of the architecture makes the profit model of the OEM diversified, which may be based on the electronic and electrical architecture towards central computing, and the original profit model of the OEM will be greatly broadened. As car companies have a large number of mobile terminals, they will have massive data (including car body data, environmental data, driving data and all kinds of data of people in the car) in the future, and they can directly reach users in the whole life cycle, from which various business models can be derived, such as software algorithms, virtual drivers, travel services, operating platforms, after-sales service and diagnosis. In the longer term, after the emergence of unmanned driving, the software ecology of vehicles has a broader imagination space. At present, some vehicle brands have been innovating the cockpit when the vehicle is stationary, in order to stimulate and meet the increasing demands of entertainment, rest and other kinds, which also makes the vehicle go beyond the meaning of pure physical movement, similar to that of smart phones that have long gone beyond the meaning of pure communication. There are 22 kinds of games built into Tesla cars, and the technical department is working hard to introduce the game library on steam into its vehicles. In the future, Tesla cars will support the smooth running of steam. In terms of hardware, all Tesla vehicles will be equipped with AMD Ryzen chipset in 2022, which is comparable in performance to the latest Sony game console Playstation5. With the increasingly rich content ecology, cars may participate in the content sharing in the future.This may become a huge source of income. In Volkswagen’s 2030 NEW AUTO strategy, Volkswagen described the future car and its application scenarios: the future car will develop into a retreat space, a mobile office, a home travel salon or a place to rest. Due to technological progress, the car will gradually lose its negative attributes (accidents, pollution, etc.) and it will become a more popular personal travel mode than it is now. Volkswagen estimates that the auto market will reach 5 trillion euros in 2030, ten times the size of the current smartphone market, mainly due to the improvement of software and autonomous driving service capabilities. Volkswagen will form a new business model in the new future of the automobile industry. The profit pool consists of vehicle hardware, software, batteries, charging and mobile travel solutions. Volkswagen believes that cars will still be personalized products in the future (thinking that customers still need differentiated car shapes, brands and services), but compared with the traditional car era, brand differences will come more from software and services. In addition to vehicle sales and vehicle platform sales, the future profit pool of Volkswagen plans to include software outsourcing, battery and energy supplement services, and travel solutions (algorithm outsourcing, mobile services). Facing consumers, services can be activated by paying as needed. Before that, Volkswagen will complete the unification of hardware platform (namely SSP platform) and unified software architecture (electrical architecture E3 2.0 +VW.OS).
Pay attention to WeChat official account, click "……" in the upper right corner of WeChat official account’s homepage, set a star sign, and pay attention to the latest information of smart car electronics and software in real time.