Source of this article: Times Weekly Author: Wu Kai

"Hey, ideal classmate, help me draw a landscape painting."

From the earliest voice-controlled windows, screen display animations, and cruise control, to the current voice call for most components inside and outside the car, generative in-vehicle systems, and end-to-end intelligent driving...... The development and application of intelligent technology is making electric vehicles full of possibilities.

In a short period of time, electric vehicles have become a second "home" for many users. And this "home" seems to be very powerful, studying, entertainment, traveling, and resting are all indispensable. From the user's point of view, it may be difficult for them to fully understand why electric vehicles are so smart and intimate, but in the eyes of "Li Auto", all this is inseparable from a car system.

The car production system is like the production system on the mobile phone, which controls and exerts the functions of the vehicle through the combination of software and hardware. In the era of fuel vehicles, the combination of software and hardware of vehicles is low, and the expandable functions of vehicles are also very limited. However, in the era of electric vehicles, especially smart electric vehicles, software has given vehicles more room to play, and the automotive system has gradually become the core component of the vehicle.

The automobile production system is also an important layout direction of major mainstream car companies. For example, the "drawing" function of the ideal student is realized on the basis of the automobile system. At present, Transwarp, Li Auto's self-developed automotive system, Transwarp OS, has opened up the vehicle control domain and the intelligent driving domain, which can meet the diverse needs of vehicle intelligence, and can be called the industry's most advanced automotive system.

而就是這一領先的汽車操作系統，在3月27日，被理想汽車董事長兼CEO李想宣佈全面開源，向全行業提供。理想汽車也因此成了全球首個將汽車操作系統這一核心競爭力開放給外界的企業。

Don't want to be "stuck"

The R&D of Li Auto's automotive system began in 2020 years, but the direct cause was the global chip shortage in 0 years.

Since the second half of 2020, many car companies have faced a structural shortage of on-board chips, with chip prices skyrocketing and delivery cycles seriously lagging behind. At that time, Li Auto applied the AUTOSAR automotive system, but the chips that could be obtained to support the system were extremely limited, and the temporary replacement of other chips would greatly extend the verification cycle and cause a loss of tens of billions of yuan.

In order to get rid of the limitations of the chip shortage, Li Auto was determined to replace the car's production system. Compared with using a third-party system or a temporary system, Li Auto chose to start from scratch and develop its own automotive system.

因此在2021年，理想汽車啟動了這一自研專案，並逐步從零開始完成了方案選型、架構設計和落地，反覆運算了上百份技術方案設計文檔。為了真正實現自研汽車操作系統的落地，理想投入了兩百人的研發團隊和累計超過十億的研發費用。終於在2024年，理想汽車實現自研操作系統上車。

The automotive system in the era of intelligent electric vehicles is very different from the previous system, and its purpose is not simply to call a few vehicle functions. For example, Li Auto's self-developed automotive production system includes vehicle control operation system, intelligent driving operation system, communication middleware, virtualization platform and various types of R&D tools. In layman's terms, this automotive system is quite similar to the central nervous system of the human body.

Among them, the vehicle control system is equivalent to the cerebellum, which is responsible for the control of vehicle limbs to quickly execute various basic control commands of the vehicle; The intelligent driving system is equivalent to the brain, which processes complex thinking processes and can ensure the efficient and smooth work of autonomous driving. Communication middleware is the nervous system, which is responsible for transmitting signals and control, and supports high-speed communication between various modules in the vehicle.

The automotive system, which integrates vehicle control and intelligent driving, can greatly improve the intelligence level of the vehicle. Especially with the blessing of artificial intelligence technology, such cars are developing towards intelligent space robots. Therefore, this has become the exploration direction of many leading car companies and technology.

Take one step and look at three steps

The original intention of Li Auto's self-developed automotive system is to reduce its dependence on corresponding chips. However, in the design process, Li Auto added a lot of ingenuity, making this automotive system a smart car system that can be used for reference and sharing in the industry.

On the one hand, Li Auto's automotive system is flexible and can support more hardware.

Ideal Transwarp OS realizes the decoupling of software and hardware, and can complete chip adaptation and verification within four weeks at the earliest, saving up to 5 months compared with using AUTOSAR. The system supports all automotive chip architectures and chips on the market, greatly reducing the dependence on specific chips.

On the other hand, Li Auto's automotive system performance is stronger.

Through cross-system architecture design, Li Auto enables system cooperation between multiple controllers to achieve deep integration of the whole link of perception, decision-making, and execution, and global collaborative optimization. This makes it 7 times faster and 0 times more stable than AUTOSAR. For example, in the high-speed driving state of 0km/h, the AEB automatic emergency braking distance of Li Auto can be shortened by 0 meters.

In terms of vehicle data security, this system has a three-layer security system of "one cloud, one vehicle, one controller". Compared with the AUTOSAR software implementation mode, its response speed is 4 times faster, and it can realize full-link device authentication, access control, and data encryption. Based on this, the vehicle system can effectively eliminate the risk of privacy leakage and realize the information security protection of the vehicle domain.

While it is greatly improved compared with the AUTOSAR system in many aspects, the cost of Ideal Transwarp OS is lower.

Previously, with the intellectualization of the whole vehicle, the functions of the vehicle are growing day by day, and each functional domain has independent sensors and controllers, which is a major test for the hardware cost and computing power cost of the vehicle. Li Auto's operating system can share AI computing power and hardware equipment, reducing the implementation cost of each function.

On the whole, through the innovation of the self-developed production system, Li Auto has reduced the BOM cost (bill of materials cost) by billions of yuan every year, and the iterative update of the software has become more efficient, which is a capability that AUTOSAR does not have. As a result, the remaining costs are fed back to the consumer.

Li Auto's automotive system also has the advantage of a flexible and customizable mechanism, that is, it realizes technical transparency through open source code, allowing developers to deeply participate in architecture optimization and function customization. This system also supports various services from the vehicle control domain to the intelligent driving domain, and can provide flexible and tailorable modular services. This opens up the possibility for its open source.

Want to be the Android of the automotive industry?

With such a leading automobile system, why should Li Auto "give it away"?

At present, mainstream car companies are competing for intelligence, but smart cars do not have a professional system. At the same time, each car company operates in its own way, ignoring industrial collaboration and vertical joint innovation, resulting in no vehicle-oriented production system in the open source ecosystem. Finally, and most intuitively, the research and development of the system is complex and extremely consuming of human and financial resources, but it is difficult to achieve economies of scale for the system developed by a single enterprise in closed source.

Based on these three conclusions, Li Auto decided to open source Transwarp OS to build an end-to-end distributed system with high performance, high safety and low resource consumption for autonomous driving, and reconstruct the industry innovation and cooperation ecosystem. This is also the fundamental thinking of Li Auto to open source the automobile system.

Behind this, in fact, is the corporate responsibility of Li Auto.

Li Auto is a leader among China's own brand car companies, and its product intelligence level is leading the industry. For example, at the level of intelligent interaction, Ideal Auto, which has Ideal Classmate and MindGPT (Li Auto's fully self-developed multi-modal cognitive model), can achieve rapid, accurate, continuous, and personalized voice interaction and control functions, and can generate content in real time according to needs.

理想汽車的智能駕駛也在快速反覆運算進步。去年11月，理想汽車“車位到車位”智能駕駛功能就已上線，車輛能夠自動出入車位、上下匝道、繞行環島等。更不用說主流的自動泊車、高快領航、紅綠燈識別功能。

All of the above are realized on the self-developed automobile system of Li Auto. Now, the leader of smart electric vehicles has decided to share their thinking, wisdom and achievements in the automotive system with those who come after them.

So what is the significance of Li Auto's move?

In fact, the automotive system is the foundation of vehicle intelligence, and it is the foundation for car companies to master the high-end intelligent body in the AGI (general artificial intelligence) era.

At present, the automobile is changing from a simple means of transportation to an intelligent mobile terminal, and the automobile industry will usher in the era of AGI, so an automobile system that can manage the resources of the whole vehicle with the highest efficiency and strongest performance is the center of the second half of the intelligent automobile industry, and it is also the foundation for high-order intelligent agents.

For enterprises, Li Auto's self-developed automotive production system means that it can break the long-term technology monopoly and avoid the risk of "stuck neck"; At the same time, it can help car companies realize chip freedom and drive the rise of domestic chips and even local supply chains, which is of important demonstration significance for industrial innovation and development.

For industry collaboration and linkage innovation, system open source can also avoid repeated "wheel building" for the industry, help enterprises reduce basic R&D investment, and avoid waste of resources. Perhaps after this open source, China's innovative technologies will join hands to go global and build an "innovation community".

As the old saying goes, "if you reach it, you will help the world". In fact, Li Auto, which has been established less than 10 years ago, has only just gained a firm foothold, but it has already regarded the healthy development of the industry as its own responsibility. From this point of view, Li Auto, as a latecomer in the industry, has shown more responsibility.