Who killed the deer is unknown.
Author | Zhu Shiyun
Without waiting for suppliers, the second German car company came up with a specific self-research plan for the vehicle OS system.
On February 23, Mercedes-Benz (Mercedes-Benz) announced that its own vehicle operating system MB.OS, based on the domain centralized electronic and electrical architecture (E/E), will be introduced to the market with MMA platform models in 2025.
Based on MB· OS, Mercedes-Benz will cooperate with Google, NVIDIA, AutoNavi Map, and Tencent Cloud to improve the infotainment system experience, and is committed to increasing the application speed of L3 autonomous driving system from the current 60km/h to 130km/h.
In 2019, the Volkswagen Group announced the establishment of "Car.Software" to create "VW. The industry thought VW.OS would be the beginning of "taking out Tesla." But three years have passed, ID. series car machine black screen, Diess class, VW. The overall postponement of OS launch to 2025 shows the huge challenges faced by traditional car companies in the software field.
Therefore, when the new car forces have "played" with domain control and OS for three years, the models equipped with domain control E/E architecture and vehicle OS of China's mainstream car companies will be widely implemented in 2023. For MB.OS, Mercedes-Benz executives are frequently asked the question: How does Mercedes-Benz deal with competition from Chinese rivals?
01
OS, the cornerstone technology of smart cars
"Own vehicle operating system is the only way to maintain the digital sovereignty of the brain and central nervous system of future vehicles." Mercedes-Benz CEO Kang Linsong explained in an interview with the media in 2020 why he developed MB.OS.
In 2021, SAIC Chairman Chen Hong was put on the cusp of public opinion for a similar "soul theory", and then SAIC Software Company Zero Beam Technology released the Galaxy "Intelligent Car Full Stack Solution 1.0" including the operating system;
At the end of that year, the M5, which applied Huawei's HarmonyOS system, quickly became a dark horse after it was launched;
In 2023, Chinese car companies have shown that there is almost "one OS" in their plans.
What exactly is OS (Operating System)? Why is it so important?
Android and IOS are the most familiar OS, but to be precise, they belong to the generalized operating system: downward is strongly related to hardware computing platforms such as SOC, upward facing developers, to achieve soft and hard decoupling (the mouse can point the position can also be used as a brush, the flash can be taken pictures can also be a flashlight), and a collection of software that supports a series of system service development.
Generalized operating systems are overwhelmingly based on kernel development. For example, Android is based on Linux, IOS is based on Unix, and the kernel is a narrow operating system, providing basic functions such as memory, files, CPU scheduling, and input and output management.
In the past, traditional cars "don't use" generalized OS.
Traditional automobiles use a distributed E/E architecture, with one ECU (on-board computer) for one function, running basic software programs or simple embedded real-time OS (OS designed for specific applications). Therefore, a function independently has a set of software and hardware resources to ensure the high reliability of the function.
However, in the era of intelligent vehicles, to carry out intelligent driving + voice assistant chat + vehicle at the same time, in order to realize the automatic closing of the window of the sprinkler truck in front of it, it is necessary to control and manage the software and hardware resources of the whole vehicle, rather than a single function;
It is also necessary to realize the decoupling of software and hardware to release the various potentials of hardware (such as cameras can be used for ADAS, can take photos and retouch and upload moments of friends), and support the development of upper-layer applications——
Centralized E/E architecture and generalized OS emerged on demand and became the cornerstone technology of automotive intelligence.
02
Car OS, the same starting line between China and abroad
"The road to a fully sustainable energy for the earth" is one of the few Musk gave for this investor conference.
In the cornerstone technology field of automotive generalized operating system (referred to as automotive OS), Chinese car companies rarely stand on the same starting line as Mercedes-Benz and other century-old multinational car companies.
In the era of hard-soft coupling ECUs, OEMs can directly purchase ECUs with this function from suppliers and add them to the vehicle system for debugging and adaptation.
However, as the electrification of vehicles continues to increase, it becomes very difficult to integrate ECUs with different "races, cultures, and languages" into one system, and development costs and cycles increase exponentially.
In 2003, nine OEMs and Tier 1 suppliers, including BMW, Bosch, Continental and Mercedes-Benz, formed an open alliance to build a standardized platform for communication interfaces, handshake protocols, data types, configuration functions and other contents between different ECU software - AUTOSAR (Automotive Open System Architecture).
As a core member of AUTOSAR, international car companies such as Mercedes-Benz have naturally accumulated more application experience than Chinese car companies.
However, AUTOSAR is essentially the middleware part of the automotive OS, and it is defined and formed for the distributed architecture, resulting in its functional inability to support the core requirements of intelligent vehicles such as vehicle OTA and intelligent driving on centralized E/E architectures.
More importantly, the system architecture design of divide and conquer is very difficult for developers who want to call software and hardware resources across systems and even domains to produce new functions, and it is difficult to form new applications and ecosystems to realize the fundamental model transformation from functional cars to intelligent vehicles.
There are no off-the-shelf products available, so I have to do it myself.
Based on QNX, Linux, Android three narrow sense OS (kernel), Tesla Version, Volkswagen VW.OS, Mercedes-Benz MB.OS, Huawei Hongmeng OS, AliOS, NIO OS, Xpeng Xmart OS, which are independently developed enterprises' own car OS, have been unveiled one after another.
These OSs strive to achieve hardware and software decoupling and atomic function packaging in automobiles, providing application developers with a basic platform that is easy to develop.
It is worth noting that HarmonyOS is currently in the stage of multi-core coexistence of Linux, LiteOS and HarmonyOS kernel, and will return to the main body of HarmonyOS in the future to truly benchmark the Android system from the bottom.
03
Race Point One: Start at different depths
From the technical starting point, although they are all customized development based on the kernel, the customization depth of each system is different.
Tesla, Volkswagen, and Mercedes-Benz take the RAM-type route (Random Access Memory, which can be read and written at any time), and carry out in-depth customized development such as modifying the kernel, hardware drivers, runtime environment, application framework, etc. on top of the narrow OS;
Most domestic manufacturers such as NIO and Xpeng take the ROM-type route (read-only memory, Read-Only Memory, can only read out information that cannot be written), does not involve changing the kernel, and generally only modifies the applications that come with the operating system.
In RAM mode, the modification of the narrow OS can make the generalized OS more suitable for the characteristics of the car scene, bring more flexibility, and cover a wider range of vehicle interiors.
For example, VW.OS can run multiple underlying systems (such as Linux, QNX, Vxworks) and eventually support 12 brand models; MB. The OS was developed based on QNX in the more secure part, and Linux was responsible for the infotainment system.
Depth-wise, MB. The OS has infrastructure for hardware such as sensors and engines; The software base layer has basic functions such as security diagnostics; The middleware layer has applications and improvements to AUTOSAR; The application layer helps developers achieve software reuse through container technology; The connectivity layer ensures that MB.OS can quickly and easily connect with different external content;
Breadth-wise, MB. OS covers the technology stack from chip to cloud, comprehensively opening up vehicle functions such as infotainment, autonomous driving, body and comfort, and driving and charging. Xue Fuming, a member of the board of directors and chief technology officer of Mercedes-Benz, responsible for research and development and procurement, told Cybercar.
In contrast, ROM mode is a redevelopment of the underlying operating system, and the applicable boundary and the degree of software and hardware decoupling will be limited to a certain extent.
For example, Xmart OS covers the intelligent cabin and driving domains on the Xpeng P7, and the vehicle control domain and central gateway are not yet involved;
NIO OS cannot be iteratively universal on the two generations of NIO platforms, NT 1.0 runs Aspen Topol, NT 2.0 runs Banyan Rong. NIO will maintain the Aspen system for 15 years, bringing the cost of a software team of 500 people.
Automotive OS starting from RAM mode will have greater compatibility and flexibility, providing an environment for sustainable growth of vehicles.
As Xue Fuming said, based on MB.OS, each domain of the Mercedes-Benz MMA platform "will continue to be equipped with the most advanced chip technology to achieve supercomputing power, support fast computing through ultra-high bandwidth and ultra-low latency, and ensure that energy consumption is kept to a minimum." Especially in the infotainment and autonomous driving domains. ”
04
Race Point 2: Who mass-produces first and iterates first
From the perspective of technology accumulation logic, there is no difference between automotive OS and vehicle stability and NVH, and it is still a process that highly relies on the verification of massive real scenarios after actual mass production and continuous iteration and improvement.
However, this time, Chinese car companies entered the "actual battle" earlier.
"As of 2018, we just need to localize AUTOSAR. Because Europe was several years ahead of us before, it is definitely enough for us to use it directly (after localization) after European development. Li Bing, chief scientist of the first batch of Chinese AUTOSAR senior member units, Neusoft Reach Tips, told Cybercar.
From 2019 to 2020, as China's new car manufacturers take the lead in the field of a new generation of centralized electronic and electrical architecture, AUTOSAR as a standard can no longer meet domestic demand.
All parties in China's automotive industry chain, including Neusoft Reach Tips, began to break through AUTOSAR, explore in the "no man's land", and gradually build an automotive OS system related to centralized domain control.
For example, in 2020, the narrow OS system of the intelligent driving domain controller equipped with the Xpeng P7 is based on QNX, but its middleware is provided by the Chinese supplier Desay SV; Neusoft Reach fully supported SOA on NeuSAR 3.0 in the same year. In the field of intelligent cockpits, many self-developed OS systems have emerged.
In the same period, only Tesla and Volkswagen used centralized E/E in foreign OEMs, and the ID.3 that came out in 2019 is still a distributed E/E architecture.
In 2023, FAW, SAIC, GAC, Changan, Great Wall and other major Chinese car companies will begin to implement more centralized E/E architectures such as central supercomputing and cross-domain convergence, and apply cross-domain and even vehicle-level automotive OS systems on them.
In contrast, the overall car OS of Volkswagen and Mercedes-Benz will be on the car in 2025.
However, it is worth noting that this year, the products of Chinese car companies applying centralized E/E architecture and automotive OS will be concentrated on a small number of flagship models; When Volkswagen and Mercedes-Benz's automotive OS is launched in 2025, it will be able to support a multi-brand and multi-model system within the group.
Achieving mass production first means that there is an opportunity to complete the system iteration first and achieve stability.
Li Bing said that the stability and reliability of automotive software ultimately depends on continuous verification and improvement in the massive scenarios of actual users' daily use of cars.
In the traditional automobile era, due to the accumulation of a large number of KNOW HOW and the continuous enrichment of simulation, bench and test ground, car companies can verify 80-90% of vehicle scenarios before delivery.
However, at present, with the gradual mass production of domain controllers using SOA architecture, the industry will form a generalized automotive operating system form with "central computing" as the core and distributed characteristics, and the extremely high complexity and huge scale of automotive software will be unprecedented, and there will be many unexpected scenarios and problems in the process of user use.
"Software needs to be continuously iterated to become more and more stable."
05
The ultimate race point: the number of users and the ecology
"Android was not the best mobile operating system back then, but it won because it had the most users and the richest ecology." An industry insider said that the same is true for automotive OS.
Starting from a deeper underlying technology, Chinese car companies have a two-year first-mover advantage in mass production, but this "race" has just begun.
It aims to build its own car OS, eventually form a universal industry standard product such as Android and IOS, and completely realize the transformation from hardware sales to software charging.
Just as important as technology is capturing developers and users. Mercedes-Benz's measures around MB.OS are also aimed at attracting more developers and end users.
In terms of expectations, although software investment will account for 25% of Mercedes-Benz's future research and development, that is, more than 2.1 billion euros/year (based on R&D expenditure in fiscal 2020), its expected revenue from selling software in 2025 is only 1 billion euros.
"From Mercedes-Benz's point of view, we are still cautious about the proportion of in-car digital services or subscription model revenue in the overall revenue in the future, especially the proportion of service revenue generated after customers purchase a car, and we still need to observe." Xue Fuming said.
This means that Mercedes-Benz will still position MB.OS as a cost center for some time to come, giving patience and resources to polish the system.
At the partner level, international Google, NVIDIA, domestic AutoNavi and Tencent Cloud have become the core partners of MB.OS.
Kang Linsong said in the interview: "To build a 'software house', you don't have to lay every brick yourself. You just need to be a good architect, take control of the big picture, put your technology partners to work, and make sure you're working with the best people. ”
Considering that Google and NVIDIA have formed a wide range of development ecosystems on PC and mobile, Mercedes-Benz and their in-depth cooperation in the underlying technology has created a shortcut for making "friends of friends".
In terms of building core applications, Mercedes-Benz plans to unveil a new generation of E-class cars at the end of this year, and the intelligent cockpit part of the new car will "preemptively" apply some functions of MB.OS, which Mercedes-Benz officially calls the "0.8 version" of MB.OS.
At the same time, Mercedes-Benz has obtained L3 autonomous driving test permits in some cities in Germany and the United States, accumulating practical experience for high-end intelligent driving.
In addition, other models other than the MMA platform can also partially use the functions of MB.OS with the adjustment of the E/E architecture and other hardware during the facelift process, thereby further expanding the scope of the system.
Mercedes-Benz certainly doesn't take China's status as its largest single market in the world.
While strengthening China's R&D strength, Mercedes-Benz will also "delegate" power to China's R&D team. Magnus Josberg, chief software officer of Mercedes-Benz, said: The relevant configuration of the infotainment system of Mercedes-Benz products in the Chinese market has been completely decided by the local Chinese team.
Lu Hanchao, R&D Director of MBUX Intelligent Cockpit and IoV of Mercedes-Benz China R&D Center, further said that another task of the Chinese R&D team in addition to a deep insight into China's special needs is to explore new functions that can be applied to China and the global market.
With its own strong capital, resource advantages, and firm investment determination, can Mercedes-Benz once again stand in a leading position in the era of intelligent vehicles? And can China's auto industry rely on its first-mover advantage and China's market advantage to quickly accumulate experience, iterate and innovate, and become the formulator of new automotive standards?
Who killed the deer is unknown.