Integrated travel and parking, a 60 billion new market for domain controller manufacturers

Since 2022, a new concept has emerged in the smart car circle-driving and parking integration, which refers to using a set of hardware equipment to realize the functions of driving and parking at the same time.

The emergence of this concept stems from the evolution of automotive electronic and electrical architecture from distributed to domain centralized. Participants include algorithm companies, chip manufacturers, domain controller solution providers and OEMs.

In order to study this hottest topic nowadays, AutoHeart launches the "Integrated Mobility and Parking" topic, hoping to clarify the latest technical routes and related solutions for integrated mobility and parking from multiple perspectives. The track that integrates driving and parking is really "rolled up". Following the launch of this function on Weilai ET7, Ideal L9, Zhiji L7 and other models, according to the latest "Mobile and Parking Integrated Industry Research Report" released by Autoheart, at least 20 companies have released smart driving that supports integrated traveling and parking at this stage. plans, most of which will be put into mass production this year and next. In the first two special articles of Automobile Heart, we discussed the functional definition and implementation path of integrated parking and parking from the perspectives of algorithm companies and chip companies respectively.

For example, in the article "The difference between travel and parking is finally made clear", Automobile Heart mentioned that Nullmax, an algorithm company with a Tesla R&D background, has polished the software platform to the extreme. Under the condition of a single SOC, it can realize the integration of travel and parking covering all scenarios and all computing power platforms. Its solution has the characteristics of deep hardware reuse, low cost and high efficiency.

In fact, in this ecosystem, there is another type of participant that is very important, that is, autonomous driving domain controller manufacturers, including Desay SV, Yihang Intelligent, Freetech, Neusoft Reichi, Heduo Technology, and Jingwei Hengrun And so on, playing the role of a bridge connecting algorithms and chips to the entire vehicle.

From a lower-level logic, it is precisely because of the emergence of autonomous driving domain controllers that it is possible to realize the integration of parking and driving.

01Domain controller, a "bridge" integrating transportation and parking

Autonomous driving domain controllers were born in the evolution of automotive electronic and electrical architecture from distributed to centralized.

In the past, each low-level intelligent driving function was implemented by an independent ECU, such as lane departure and traffic recognition ECU, parking assist ECU, and rear collision warning ECU.

To realize ADAS, these independent ECUs need to be "connected in series" to work together.

However, as functions are superimposed, the vehicle wiring harness becomes more and more complex, system stability and safety are reduced, and it is difficult to make the leap to higher-level intelligence. In order to solve these problems, domain controllers came into being.

Simply put, a domain controller is a regional "central brain" that can integrate originally isolated ECUs with each other for unified scheduling and control. On this basis, driving domain control and parking domain control have emerged. The two sets of domain controls each use different chips and sensors to complete the driving and parking functions.

Although this is a step further than the previous era of independent ECUs, it is still a separation of traffic and parking. It cannot deeply multiplex sensors and share computing resources in the two domains, and it is difficult to cope with more complex scenarios such as urban NoA and AVP. On the contrary, if driving assistance and parking assistance functions can be integrated simultaneously through a domain controller, it will not only greatly reduce system costs, but also help OEMs improve development efficiency, return data, and improve users' driving experience.

The Xingbo integrated domain controller is designed based on this idea. Structurally speaking, the Xingbo integrated domain controller integrates software and hardware such as the main control chip, software operating system, middleware, and application algorithms.

In addition, it also provides a wealth of interface resources to undertake functions such as input, processing, and fusion of sensor data, as well as planning, control, and decision-making for intelligent driving. The main control chip is undoubtedly the core of the integrated domain controller, and its computing power and integration directly affect the shape of the vehicle's EE architecture.

An industry insider said that when designing domain control, the heterogeneous core performance of each chip and the degree of model adaptation must be fully considered. Taking Desay SV as an example, it has launched two generations of four sets of domain controller products based on different chips, including IPU01 and IPU02 that take the cost-effective route, and IPU03 and IPU04 that take the high-performance route. in:

• 
  

• IPU02 was launched in 2019 and is based on the Texas Instruments TDA4 chip, integrating driving and parking functions; IPU03 is based on NVIDIA Mass production was achieved on the Peng G9.

Similarly, Zhixing Technology has formed three product layouts: iDC Mid, iDC High, and SuperVision.

Among them, the iDC Mid domain controller also uses the Texas Instruments TDA4 chip and will be mass-produced in September 2022. It can not only realize NOA navigation assisted driving and L2 driving functions, but also have HPA memory parking functions, 360 panoramic views, etc. Function.

Yihang Intelligent also fully exploited the potential of TDA4 chips to create a NOA integrated travel and parking solution, and in 2016 won the first 100,000-level autonomous driving Tier1 mass production project. Basic software is also an important part of the Xingbo integrated domain controller, including the underlying operating system and middleware. Its function is to facilitate the development of application software at the upper level by OEMs and algorithm companies to quickly develop and iterate its smart driving functions.

Neusoft Reichi's new generation of parking and parking integrated domain controller comes standard with its self-developed basic software product that is compatible with the latest AUTOSAR standard - NeuSAR and related development tools, equipped with an autonomous driving SOA open architecture. This software architecture supports L0-L4 level autonomous driving functional requirements.

In order to better coordinate with vehicle model function development and mass production, the software architecture also pre-installs standardized NeuSAR basic software and autonomous driving-specific middleware:

• 
  

• It can achieve efficient decoupling of software and hardware; realize a reusable, configurable, and scalable flexible deployment mechanism for upper-layer applications.

With the basic hardware platform and software platform technology of domain controllers, integrated travel and parking applications based on algorithm and data development can be deployed on them.

02 It is necessary to build an integrated domain controller to test the software and hardware level and engineering capabilities.

Although mobile and integrated domain controllers are "in the limelight", it is not that easy to do well. From a hardware perspective, the number of components and materials used in the Xingbo integrated domain controller far exceeds the number of components in any vehicle ECU in the past, which means that the probability of system errors and failures is greatly increased.

Li Lele, vice president of Desay SV, said that for this purpose, it is necessary to carry out fail-safe design for the functional safety of all hardware, especially for large computing power domain controllers, WCCA (worst case circuit) analysis and failure probability analysis are required. Corresponding backup design. Secondly, due to the physical space limitations of the vehicle, it is also extremely challenging to reduce the electromagnetic anti-interference between devices and ensure signal integrity while minimizing the size of the integrated domain controller. In addition, due to the high power of the large computing power domain controller chip, a specially designed heat dissipation solution is required to prevent it from failing due to excessive temperature.

The current mainstream solution is to introduce liquid cooling pipelines. The challenge lies in how to monitor the temperature and accurately control the inlet water temperature and flow rate. Otherwise, the water cooling liquid will be overcooled and cause condensation inside the controller, or the water cooling liquid will be overheated and fail to dissipate heat. effect. In terms of software, there are also many areas for improvement.

The operating system is one of them. Since there are many modules in the macro kernel, loading too many modules will affect the startup time of the system. Modules that do not require high real-time performance need to be moved, such as from the kernel mode to the user mode. In addition, although Xingbo integrated domain controller manufacturers are all promoting SOA servitization, according to Liu Wei, deputy general manager of Neusoft Ruichi, the differences between each are very big.

In an interview with "Nine Chapters Smart Driving", he said that there is currently no conclusion in the industry as to whether signals can be service-oriented.

"Because, first, it depends on the entire application requirements; second, it depends on the entire vehicle architecture; third, it also depends on the computing performance of the entire hardware." In addition to focusing on autonomous driving, the integrated driving and parking domain In terms of software architecture, controller manufacturers also need to consider data upload and information security, as well as local and cloud collaborative data processing.