Huawei Cai Jianyong: Layered decoupling, win-win cooperation, and prosperity of the intelligent vehicle ecosystem

【EV Vision Report】On December 21, Huawei held the "2021 Huawei Intelligent Vehicle Solution Ecological Forum" in the form of online live broadcast, and Cai Jianyong, general manager of the intelligent car control field of Huawei Intelligent Vehicle Solution BU, said at the forum: "With the advent of software-defined cars, the complexity of software and hardware is getting higher and higher, posing a severe challenge to the automotive industry. Hierarchical decoupling, open collaboration, and win-win cooperation are the foundation for the intelligent automobile industry to become bigger and stronger. ”

In recent years, in the transformation of intelligent vehicles, there have also been a scene of "a hundred flowers blooming", there are many different directions, different evolution strategies, different development models, which one represents the future?

Cai Jianyong believes that "the stone of his mountain can attack jade", the ICT industry in the past half century, relying on information technology, with new application models, new business models as the core competitiveness, socialization and industrialization services as the value goal of emerging industries have penetrated into all fields of human activities, continue to promote and lead the rapid development of society and economy, has grown into the "fourth industry". We are familiar with computers, mobile phones, Internet, mobile networks, etc. are typical representatives of this field, the typical feature is to follow the ICT industry roughly through three stages, namely function-driven, performance-driven, intelligent and personality-driven, to computer and mobile phone as an example, the computer has experienced a large all-in-one machine, a small all-in-one machine, a microcomputer three stages

The first stage: the circuit is mainly electron tubes and transistors, a computer and a warehouse are so large, the software is mainly based on batch processing, and the software and hardware are highly bundled;

The second stage: the development of integrated circuits, greatly reduced the size of the computer, into the era of small all-in-one machines, but there are also double-door refrigerators as large, this era of operating system to UNIX-based, software and hardware began to separate, software development began to form a platform concept;

The third stage: is the development of large-scale integrated circuits, so that the size of the computer is further reduced, forming the form of desktops, notebooks, tablets we now see, this stage of hardware and software interfaces are highly standardized, ecological, shelf, users can diy according to their own needs, to meet the user's personalized needs.

Mobile phones have also experienced three eras from big brother, color screen, smart phone, big brother era of each manufacturer for the government of the software and hardware interface is highly privatized; color screen machine era, hardware interface began to standardize, unified, mobile phone manufacturers to the focus of competition to software, this stage of the operating system is blooming, basically the larger mobile phone manufacturers have their own operating system; in the era of smart phones, we see that because of the division of standards and ecology, most of the OS's competitiveness is insufficient. Finally unified to iOS and Android, at this stage os API, application ecology, hardware interface gradually unified, software can be developed and run across platforms, the industrial division of labor is reasonable, efficiency has been greatly improved.

So where is the smart car industry in? What stage does the OEM want to go to in the future? What lessons can be learned from the development of the ICT industry?

Cai Jianyong pointed out that in the past few decades of development, competition has been the key driving force to drive the continuous improvement of the complexity of the car, this trend will continue to the era of intelligent cars, and will be interpreted to the extreme in the trend of personalized development of thousands of people, these two factors will continue to drive smart cars with richer configurations, more choices, more complex combinations, and better driving experience, but at the same time introduce four major challenges to the automotive industry:

The first challenge: the configuration is facing three ten-fold growth, that is, the intelligent hardware portfolio increases by more than tenfold, the number and type of intelligent peripherals increase by more than tenfold, and the intelligent software type and code scale increase by more than tenfold, how to manage the complexity of the software and hard combinations, and balance competitiveness, quality, cost and end-to-end efficiency.

The second challenge: experience is facing three demand growth, the first is the continuous increase in user experience demand, the basic scenario is conservatively estimated to exceed 1,000 kinds; the second user experience expectations are rising, the use of smart terminals The use habits are rapidly extending to smart cars, such as OTAs; the third OEM needs to build a differentiated experience to ensure user stickiness in order to achieve good word-of-mouth communication, and how to meet the growing demand for experience while ensuring the quality of vehicle regulations.

The third challenge: in terms of supervision and regulations, first of all, the attack surface of intelligent networked vehicles is more, and secondly, the vulnerabilities increase exponentially with the size of the code by ten times and a hundred times, how to build a vehicle-level security capability mechanism to ensure that smart cars make regulators and society truly at ease. At the same time, functional safety and network security are changing from single element to multi-element coupling, and the traditional single function security, expected functional safety, and network security design have been unable to solve the security risks caused by multi-element coupling. How to ensure that smart cars truly reassure consumers and OEMs.

Despite facing so many challenges, to solve these challenges requires huge investment, but the intelligent car itself exists a large number of customized and privatized interfaces, this inefficient, wasteful traditional development model is still continuing, software customization will bring a large number of interface adaptation, drive adaptation, repeated calibration, repeated adjustment of the communication matrix and other repetitive labor, end-to-end software development efficiency is low, human resources waste is serious, in the era of traditional functional vehicles, this model can be basically maintained, but with the acceleration of vehicle intelligence, Intelligent software will show exponential growth, the transformation of software development model is imperative, and the results of hardware customization are frequent customization of hardware, customization of wiring harnesses, verification of repeated DV/PV, and repeated investment in certification. End-to-end management complexity and cost remain high, frequent line adjustments lead to wasted capacity, and model switching leads to a linear increase in total spare parts inventory.

Cai Jianyong believes that to solve the industrial challenges mentioned above, smart cars need to return to the original intention of car service to people, need a sustainable evolutionary architecture for smart cars, and the design concept should be people-oriented, around the continuous innovation of user experience and OEM business success, to have the key characteristics of "four cans":

First: the software can be upgraded, the software can be reused across models, platforms, and car companies, supporting rapid development and continuous release;

Second: the hardware is scalable, the hardware is scalable and replaceable, and peripherals such as sensors can be plugged and played;

Third: security can be guaranteed, build a vehicle-level digital security and defense-in-depth system, and support Mail Operational;

Finally, commercial sustainability makes the car a sustainable value preservation and appreciation platform, with the best return on investment throughout the life cycle.

In order to achieve the above goal of sustainable evolution of smart vehicles, the industry needs to work more closely together to reduce industrial fragmentation and inefficient duplicate inputs and cost inputs by giving full play to China's standardized scale supply advantages. In December 2020, Huawei joined the SDV Working Group of the China Association of Automobile Manufacturers, with the goal of working with automakers to promote the standardization of software and hardware interfaces for intelligent vehicles, reduce the complexity of intelligent vehicle research and development, accelerate the development process of intelligent vehicles, and work with upstream and downstream enterprises to expand and strengthen China's intelligent automobile industry chain. Based on the hierarchical architecture of software-defined automotive software and hardware decoupling, the SDV working group advocates a reasonable division of labor in the industry, based on a unified standardized API, and each enterprise gives full play to its own advantages and focuses on product innovation and research and development in its own field. At present, the SDV working group has more than 70 member units to join, covering 20 OEMs and more than 50 upstream and downstream enterprises, including domestic and foreign parts suppliers, automotive application software development companies, basic platform providers, automotive engineering companies, etc. Focusing on the work goals of the SDV working group, after more than a year of joint efforts, we have released the software-defined automotive atomic service API and device abstract API, including body control, thermal management API official draft 1.0, motion control, energy management, chassis domain, power domain draft for comment.

The formulation process of API reference follows the principle of open collaboration, through the standardized API definition, review, decision-making and release process, fully gather industrial wisdom and experience, OEMs, parts suppliers, industry organizations, platform suppliers, software companies and other multi-party units, through different channels to deeply participate in API definition, through the SDV working group released by the API after full discussion in the industry, and on-demand iterative update, so as to ensure that the API is easy to use and easy to use. Since the first plenary meeting of the SDV working group in April 2021, the working group has organized two full-staff exchange meetings, summarized nearly 50 FAQs, more than 30 group definition seminars, absorbed more than 130 offline review opinions, and more than ten partners have truly participated in API reference implementation and integration verification. Through the continuous efforts of the SDV working group, the entire API definition process is continuously standardized, and the API definition work is also carried out in an orderly manner.

Huawei will continue to work with SDV working group members to improve the SDV API specification around the hierarchical decoupling architecture, and in Q2 2021, the body domain and thermal management domain atomic service device abstract API draft was released, and Q3 officially released the version of these two fields; in October 2021, the energy management, motion control atomic service, and the device abstract draft of the power domain and chassis domain were released, and the working group member units are currently organizing discussion and review. Q1 2022 will officially release APIs in these areas; Q2 2022 plans to release a version of intelligent services 1.0, including image recognition services, perception services, positioning services, and supporting the release of abstract APIs for various sensor devices, such as cameras, millimeter-wave radar, lidar, etc. And for the high-end body domain function, supporting the release of version 2.0, such as electric doors, zero gravity seats, follow-up headlights and so on. In Q4 2022, we will continue to update the existing released domain capabilities and launch the 2.0 version, such as suspension control, wireless charging, car charging, grille control, etc.

Huawei has created a visual API design and management tool, hoping to help the industry gather wisdom and strength to build a high-quality API interface to prosper the intelligent car ecosystem, this tool has the characteristics of visualization, lightweight and intelligence:

First: When consulting the API, the device abstraction and atomic services can display the API interface document in full and in full version, and can also be easily consulted and generated

Second: when modifying the API, it can be dragged and dropped, graphical display, simple and efficient, and can be compatible with different versions of the AUTOSAR standard

Third: API version traceability, the tool can intelligently track and compare API versions, one-click view of all the difference points of the API, so that all OEMs and parts manufacturers using these APIs can easily trace the differences in the API.