The long-closed intelligent driving system was cracked at the Yunqi Conference

Jia Haonan from Aofei Temple
Quantum Bit Report | Public Account QbitAI

Intelligent driving operating system , the most basic but the most important.

It is the most critical core of a complete vehicle system and the heart of the next generation of smart cars. Chips and intelligent driving algorithm applications all rely on it for connection.

More than 90% of the global vehicle control OS market is dominated by the foreign closed-source system QNX, and the domestic market has never had any self-developed mass-produced products.

But a crack was knocked into this situation.

At the Yunqi Conference, Banma Smart Driving just released its self-developed AliOS intelligent driving system core with the world's highest product safety certification (ISO 26262 ASIL-D) .

It is officially announced that it will be released next year, solving the problem of developing a self-developed vehicle control system from scratch.

As a new player in the vehicle control operating system, Banma Smart Driving has also brought a gameplay that may change the industry: opening the vehicle control system kernel to car companies for free.

What is the core of the AliOS intelligent driving system?

What exactly is a vehicle control operating system?

It supports all driving-related functions such as intelligent driving algorithms, body control modules, wire-controlled chassis, etc. It is the nerve and bridge connecting various functions and computing chips and functional modules.

Ordinary users do not have direct contact with it on weekdays, but they play a key role all the time.

The vehicle control operating system and the intelligent interaction system together constitute what we usually call the car operating system.

As for the vehicle control system, it is necessary to clarify the difference between "time-sharing operating system" and "real-time operating system".

The former divides the system processing time and memory space according to time, and switches them to different users in turn, so that they operate independently and do not interfere with each other.

The latter emphasizes rapid response to emergencies, completing processing operations within a specified time, and allocating CPU resources according to task priority. Such characteristics are exactly what is needed in driving scenarios.

Requirements such as emergency braking and rapid response determine that the vehicle control system must be real-time online and meet high reliability.

The AliOS intelligent driving system kernel released by Zebra Smart Driving is the core part of the vehicle control operating system.

Among them, hard real-time means that the operation must be completed within the specified time, which is guaranteed at the operating system design level. The response cycle is upgraded from the traditional hundreds of milliseconds to 10 milliseconds, bringing exponential improvements in both control and safety.

Reliability mainly comes from the microkernel architecture , which supports CPU, memory isolation, and address isolation between processes. The driver and file system are placed in the user space, so that the failure of a single module will not affect the entire core.

To upgrade the system under the microkernel architecture, only a single module needs to be replaced without changing the entire operating system, which greatly enhances scalability and OTA efficiency.

The AliOS intelligent driving system kernel also has a deeper level of openness, which complies with the API standards released by the Institute of Electrical and Electronics Engineers (IEEE) and is supported by the industry's mainstream operating systems QNX, Unix, Linux, and MAC OS.

In addition, it supports the SOA concept at the software application level and supports multi-core chip (SMP/AMP) architecture at the hardware communication level.

This will enable the decoupling of software and hardware in smart cars, with fast software OTA and plug-and-play hardware, without interfering with each other. The popular concept of "software-defined car" in the industry is actually this at the technical level.

After understanding the technical features of the core of the AliOS intelligent driving system, you can actually understand that this is actually Banma Zhixing "laying the foundation" for the domestic in-vehicle operating system.

What are the characteristics of the AliOS intelligent driving system core?

The vehicle control system is fundamentally different from the general operating system. From the perspective of safety, it needs to be more stringent in terms of indicators such as latency and reliability.

The AliOS intelligent driving system proposes five iconic indicators: high real-time performance, high security, high reliability, high functional safety, and high performance computing.

High real-time performance refers to the 10ms response cycle introduced earlier.

It is worth mentioning that the 10ms response level of the in-vehicle OS even exceeds that of daily-used mobile phones or PCs. This is also the fundamental difference between the safety and latency requirements of automobiles and consumer electronics.

The in-vehicle system must be safe, which means not only fast response but also high reliability.

The losses caused by personal users' mobile phones and computers going down are acceptable, but the system downtime of private cars is a matter of life and death.

The reliability of the AliOS intelligent driving system is reflected in its microkernel structure. The failure of one module will not affect the entire core.

Second, it is required to be able to operate stably for a long time, and the system functions and services provided during operation should remain available. Among them, key vehicle control equipment (such as brakes) must meet the highest safety standards.

This is in stark contrast to general-purpose operating systems that allow freezes, reboots, and partial functional failures.

The AliOS intelligent driving system focuses on safety, which is directly reflected in the highest ASIL D standard.

ASIL (Automotive Safety Integration Level) is a risk level used in ISO 26262 and GBT34590 standards to assess the hazards that may be caused by functional failures of road vehicles.

ASIL D is the highest level, and ASIL A is the lowest. The higher the ASIL level, the more complex the system design, the more development cycle, and the more development costs increase exponentially, which means that a higher price must be paid to achieve safety. Why do you say that?

The first step is to obtain the ASIL "ISO 26262 Automotive Functional Safety Management System Certification Certificate", which indicates that the software development covers the entire process of demand planning, design, implementation, integration, verification, confirmation, configuration, and manufacturing, and has met the ASIL D safety level requirements of the automotive industry.

The second is to obtain product-level functional safety certification (ISO 26262 ASIL-D). ASIL D certification means that the autonomous driving operating system kernel can meet relevant safety requirements.

This poses a huge challenge to the company's technology reserves, talent, management system and financial resources.

QNX, which has monopolized the market, has accumulated 40 years of experience, and in China, only Banma Intelligent Driving, incubated by Alibaba, and Huawei Hongmeng are currently doing this.

Finally, there is high performance computing.

The AliOS intelligent driving system kernel can provide powerful computing power for ADAS and autonomous driving to meet the requirements of image recognition and decision-making calculations. Secondly, it has a strong data throughput capacity to meet the real-time access and processing of multi-sensor data. Third, it has a high degree of flexibility/scalability/programmability to meet the needs of various algorithm models. Finally, it has fast learning and ease of use to meet the debugging, tuning, and testing required for ADAS and autonomous driving algorithms.

Having said so much, when can we see it?

Officials revealed that smart cars using the AliOS intelligent driving system core will be available in 2022 .

Is it necessary to develop your own car control OS?

Ten years ago, the free whirlwind of the Android system reduced the manufacturing cost of smartphones, promoted the rise of the mobile Internet era, and opened up the smartphone market for Samsung, Huawei, Xiaomi, and vivo to compete with the Apple camp.

This scene seems familiar.

Because this time Banma Smart Driving also announced its strategy of opening the AliOS intelligent driving system to users for free.

Entering the market with an open attitude, Banma Smart Driving has a clear idea:

"Free" can reduce the software cost of new car manufacturing, attract more upstream and downstream partners to join, and accelerate the development of "software-defined cars".

The more partners participate in "software-defined cars", the faster the industry will develop, and the more solid Zebra's "foundation" will be.

The next step to lay a solid foundation is naturally to move towards a complete in-vehicle operating system, which is also the ultimate goal of Zebra Smart Driving's "trilogy" strategic planning.

In fact, Banma has already made some achievements in the smart cockpit system , another part of the in-vehicle operating system .

The current installed capacity has exceeded 2 million, making it a "little giant" in the industry.

However, only Zebra Smart Driving and Huawei Hongmeng are currently advancing the complete in-vehicle operating system that aims at the highest safety standard ASIL D.

Strength is one thing, the key is why?

At the Global New Energy and Smart Vehicle Supply Chain Innovation Conference held in early October, some experts issued an urgent call, hoping that domestic companies would make every effort to seize the commanding heights of automotive operating systems.

First of all, in terms of user experience, the multi-screen interaction experience of QNX system or Android is not good, but Zebra Smart Driving can achieve the display boundary fusion of each application and communication between applications. This is because Zebra Smart Driving designs the operating system from the bottom up.

Secondly, there are risks in terms of industrial security. Foreign commercial closed-source operating systems are at risk of having their authorization terminated. On the other hand, the safety of the entire vehicle in the future will inevitably include data security, and self-development is also the best choice.

In addition to security, the underlying operating system also plays a key role in optimizing and adapting chips, and developing applications and algorithms on the connector layer.

The healthy development of the smart car ecosystem cannot be without this key link that connects the past and the future.

Therefore, the birth of the AliOS intelligent driving system core is also accompanied by a free and open strategy and two industry collaborative alliances.

One is the "New Automotive Cloud Laboratory" established with the participation of industry partners and top universities, and the other is the "New Automotive Chip Alliance" participated by semiconductor companies.

Zebra Smart Driving emphasized that "Android of the car industry" is not its goal, so what is the significance of its entry into the difficult and challenging operating system track?

This is actually easy to understand.

Today, in addition to the QNX+Android in-vehicle system, Zebra Smart Driving provides another option with lower cost, better performance, and a more open ecosystem. The key is that it has obvious advantages in safety and controllability.

Isn’t this the greatest significance?