Will 2024 be the key window period for cabin-driving integration?

As the electronic and electrical architecture of the vehicle evolves, centralization is the general trend.

Thanks to the substantial increase in the terminal market's demand for intelligence, as well as the intense involution of the automotive industry to reduce costs and increase efficiency, and the industry continues to make new breakthroughs in high-performance computing chips, in the past two years, both smart driving and smart cabins have The core sector has basically achieved intra-regional integration.

According to configuration data from the Gasgoo Automotive Research Institute , in 2023, the delivery volume of standard cockpit domain control front-end equipment in the Chinese market will reach 3.476 million units, and the adoption rate will increase to 16.5%; the delivery volume of smart driving domain control front-end equipment will be 1.839 million units, and the implementation rate will increase. to 8.7%.

At present, as mainstream car companies are rushing to promote the research and development and implementation of centralized electronic and electrical architectures, the demand for cross-domain integration between various functional areas has been further increased. Against this background, cabin-driving integration is a key transition stage. It has become a new competitive point in the industry.

Cabin and driver merge, the battle begins

The so-called cabin-driving integration, as the name suggests, is to highly integrate the cockpit domain and the intelligent driving domain into a high-performance computing unit to achieve a comprehensive integration of hardware, software and applications, thereby better supporting the deployment and update of new functions, improving user experience, and at the same time Shorten the development cycle and reduce vehicle costs.

Image source: Gasgoo

Under the current domain-centralized architecture, although the intelligent driving domain and the cockpit domain have achieved intra-domain integration, the two domains are still independent of each other, and the whole is Two Box and Two Board. Cabin-driving integration is committed to realizing One Box, One Board, and One Chip. One Chip is generally considered to be the "ultimate solution" for cabin-driving integration, that is, the smart driving domain and the smart cabin domain can be run simultaneously on one SoC chip.

However, due to the higher level of system integration, single-chip cabin-driving fusion has higher requirements than the first two solutions in terms of software adaptation and hardware selection.

In the early days, restricted by factors such as chip performance, software technology level, and architectural solutions, cabin-driving integration mainly adopted multi-chip solutions.

For example, Leapmotor's "four-leaf clover" architecture uses "Qualcomm SA8295+NVIDIA Orin" for its high-end configuration, supporting high-speed NOA, urban NOA and mainstream smart cockpit applications. Yikatong's Super Brain uses "Xinqing Technology Longying No. 1 + Black Sesame Smart A1000" to achieve cabin-driving integration. Desay SV’s vehicle-mounted intelligent central computing platform ICP Aurora is even equipped with three high-performance SoCs, namely NVIDIA Orin, Qualcomm SA8295 and Black Sesame Huashan A1000.

The real turning point for single-chip cabin and driving integration will begin in the second half of 2022.

In September of this year, NVIDIA and Qualcomm successively launched DRIVE Thor and Snapdragon Ride™ Flex SoC for central computing architecture. The computing power of up to 2000Tops makes it possible to deploy the smart cabin domain and the smart driving domain simultaneously on one chip. Driving integration thus kicks off the development of single SoC. In the past period, many OEMs and Tier 1 have been actively developing cabin-driving integration solutions based on single chips.

Image source: Bosch

For example, Bosch's new cross-domain computing platform, which debuted globally at CES 2024, uses Qualcomm's latest generation Snapdragon Ride™ Flex SoC to support running numerous smart cockpit and smart driving functions simultaneously on a single SoC, including automatic parking, Lane detection, intelligent personalized navigation, voice assistance, driving assistance functions, etc. According to previously announced information, this chip is expected to begin mass production in 2024.

In addition, Chuangda's subsidiaries Changxing Zhijia, Meijia Technology, and Cheliantianxia also demonstrated domain controllers based on Snapdragon Ride™ Flex SoC during CES 2024, promoting the rapid development of cabin-driving integration.

It can be seen that in this round of cabin-driving integration technology competition, Qualcomm’s Snapdragon Ride™ Flex SoC has become the first choice of many Tier1s.

New car companies such as Li Auto, Jikrypton and Nezha Automobile have all confirmed that they will use NVIDIA's DRIVE Thor in future new cars. The first Jikrypton new car equipped with DRIVE Thor is scheduled to be launched in 2025.

“Because the above car companies currently use Nvidia chips in their smart driving systems, based on this, the reusability of certain underlying developments may be better compared to using products from the same chip factory. . And Qualcomm’s previous SoCs mainly focused on the cockpit, and it can be found that the Tier 1 partners it cooperated with were basically customers of its cockpit domain control in the past." Analysts from Gasgoo Automotive Research Institute said.

It is worth noting that in the past few years, Qualcomm and Nvidia have established significant dominant positions in the fields of smart cockpits and autonomous driving respectively with their deep accumulation in the chip field. Nowadays, as the trend of cabin-driving integration becomes increasingly prominent, the two leading companies have quietly launched a new round of strategic competition and technology game.

Image source: Black Sesame Intelligence

Also joining the battle is Black Sesame Intelligence. On the eve of the 2023 Shanghai Auto Show, Black Sesame Intelligence announced the official launch of the cross-domain computing chip platform "Wudang" series, as well as the series' first chip C1200, providing a "Chinese solution" for the vehicle E/E architecture to move into the central computing era.

According to Yang Yuxin, chief marketing officer of Black Sesame Intelligence, the cabin-driving integrated and single-chip NOA solution based on the Wudang series C1200 is expected to be mass-produced from the end of 2024 to the beginning of 2025. At present, C1200 has completed complete testing after tape-out, functional performance verification has been successful, and samples can be provided to customers.

In addition, Baidu, Aptiv , Visteon, Zongmu Technology and Heduo Technology have also stated that they are promoting the research and development of integrated cabin and driving. Among them, Aptiv’s cross-domain integrated computing platform is reported to be developed by the Aptiv China R&D team. It is built based on local system-level high-performance chips and covers the three control domains of intelligent cockpit, intelligent assisted driving and automatic parking.

It is foreseeable that, driven by the synergy effect of deepening the industrial chain, cabin-driving integration technology is gradually approaching the "critical point" of implementation.

2024 will be a critical window period

As soon as this year, an integrated cabin-driving solution based on a single chip with high computing power is expected to become a reality.

“Especially with the development of cabin-driving fusion chips, the rapid evolution of SOA architecture technology, and the continued growth of OEM demand for cost optimization, and the increasing user demand for intelligent functions, the superposition of multiple factors has promoted the mass production application speed of single-chip cabin-driving fusion. It is accelerating," said the relevant person in charge of Desay SV. Against this background, it is expected that 2024 will be an important year of opportunity for the implementation of single-chip cabin-driving integration.

Qian Qian, deputy director of the Central Research Institute of Hangsheng Technology Center, also believes that the future of cabin-driving integration has arrived, and more car companies are expected to gradually achieve mass production this year and next.

"The integration of cabin and driving is an external pull caused by the increasing demand for intelligence from users, and also comes from the internal drive of car companies to optimize costs. Among them, new energy vehicles will be the first to start to be implemented as a core carrier, and as new energy vehicles The continuous increase in penetration rate and the upgrading of traditional oil trucks will accelerate the evolution process of cabin-driving integration." Talking about the core driving force for the current development of cabin-driving integration, Qian Qian said.

But this process will not happen overnight. Regarding the different forms of cabin-driving integration, many people in the industry believe that One-Box, as a less difficult integration solution, is expected to be realized first; and One Board, which combines smart driving chips and cockpit chips Integrated on one core board to further improve the integration level, but there are still two SoCs, which can be regarded as another transitional solution; the last one is One Chip, which achieves true cabin-driving integration, but this solution strongly relies on Mass production breakthrough of high-performance SoC.

Image source: Gasgoo

It can also be seen from this that although cabin-driving integration has brought rich imagination space to the intelligent evolution of the entire vehicle, it is not easy to implement it.

"Cabin-drive integration is the product of the gradual innovation of electronic and electrical architecture. It is an update at the architectural level. It is not just as simple as two circuit boards turning into one circuit board, or two chips turning into one chip." Qian Qian pointed out. Therefore, there are bound to be many technical and performance challenges.