On June 28, Du Qiang, President and CTO of Neusoft Reach, was invited to attend the 2nd Infineon Automotive Innovation Summit and delivered a keynote speech on "Automotive Software Architecture and Solutions under SDV+AI", explaining that the automotive industry is facing challenges such as development complexity and cost amortization caused by software system reconstruction under the two-wheel drive change of SDV and AI. At the same time, GM's automotive software needs to have the ability to support the rapid iteration and continuous innovation of AI technology, so as to accelerate the pace of automotive intelligence and promote the prosperity and development of the automotive industry.
AI empowers new automotive intelligence to drive a wider range of innovation and continuous iteration of functions
With the take-off of AI technology driving a wide range of attempts in the automotive field, automotive intelligence is ushering in a new round of competition focus, how to use AI technology capabilities to improve product intelligence, create product differentiation, so as to truly improve the user experience, has become the direction of the current automotive industry in the new round of technology wave. However, the deep integration of AI and automobiles has brought about substantial changes, and there are still a series of challenges:
The first is the huge software scale and cost sharing problem: the intelligent transformation of automobiles has led to a surge in the scale of automotive software, and at the same time, the strict security requirements such as functional safety and information security have made the complexity of automotive software and hardware architecture much higher than that of other software industries, and the cost of automotive software development is high, and the amortization unit cost is still high.
Second, industry standards urgently need to be unified: in the Internet field, open standards such as W3C standards and HTTP protocols can support start-ups to easily carry out application innovation and business model exploration. However, the automotive industry still lacks a unified software development standard, and the high threshold of the automotive development system restricts the rapid introduction and implementation of a wider range of innovative ideas.
Third, the user stickiness of automotive AI applications has not yet been established: although the Internet has ushered in the "ChatGPT" moment, there is still a big gap between AI applications and traditional Internet applications in terms of customer stickiness, especially in the automotive field, where "killer" applications with high user stickiness have not yet been found.
Fourth, there is still a gap between the automotive development cycle and the AI iteration speed: the iteration speed of applications derived from generative AI technology far exceeds the iteration speed of the model itself, and even the iteration speed of software OTA. How to adapt to the speed of application innovation under the rapid development of AI technology puts forward higher requirements for the versatility, reliability, and compatibility of software platforms.
In the face of the multiple challenges of fierce market competition, cost pressure on a single vehicle and continuous improvement of software value, it is necessary to build a general-purpose software platform that can run on tens of millions of vehicles, effectively achieve cross-platform compatibility and development cost reduction and efficiency increase, and provide assistance and support for AI on the car, which has become the only way for the automotive industry to continue to innovate and develop.
General-purpose software platform: Achieve a high degree of software reuse and build a solid foundation for AI application innovation
Neusoft Reach hopes to build a set of general software platforms, which provide standard basic software and rich middleware, and support cross-domain, cross-model, and cross-platform product compatibility.
At the same time, Neusoft Reach is committed to promoting the formulation of industry standards, realizing the unification of SOA service architecture and interface specifications, so that car companies, suppliers, and third-party developers can realize convenient application development based on open standards, which greatly reduces software costs and makes the software ecosystem more prosperous.
At the same time, under the trend of accelerating the introduction of AI in vehicles, how to effectively support the innovation and iteration speed of AI applications in the car, so as to accelerate the upgrade of the intelligent level of automobiles, developers also need to have the following capabilities:
The ability to iterate quickly of software: The combination of AI and application scenarios is changing rapidly, and general-purpose software platforms need to have the ability to iterate on a weekly basis to support the continuous innovation and rapid iteration of intelligent applications with the support of AI technology.
Internet-based lightweight development mode: The general-purpose software platform needs to create a cloud-native lightweight development model for application developers, reduce the threshold requirements of the traditional automobile development process system, and provide a convenient and efficient development environment for application developers.
Flexible and open development ecosystem: In-vehicle AI application innovation needs to rely on an open and prosperous ecological environment, rather than a closed system, to stimulate and attract more software talents to collaborate and innovate, repeatedly polish and iterate, and finally create intelligent functions that truly meet user needs.
Vehicle-cloud synergy: It is difficult for a single hardware to support the continuous evolution of AI technology, so the introduction of vehicle-cloud integrated architecture design from the beginning of application development to achieve seamless connection between the vehicle and the cloud and the separation of computing will support the continuous provision of excellent user experience throughout the vehicle lifecycle.
Neusoft Reach NeuSAR is on the verge of replicating every line of code to tens of millions of vehicles
Based on years of practice and experience accumulation in software architecture and development methods in technology R&D and mass production projects, Neusoft Reach's self-developed software development platform NeuSAR not only provides basic software that meets the latest AUTOSAR standards, but also builds middleware components that meet cross-domain integration, autonomous driving, and vehicle-cloud integration, realizing flexible deployment of applications across vehicles and clouds, and providing assistance for efficient and convenient development of AI applications.
NeuSAR cCore/aCore is AUTOSAR R21-11 compliant, provides reliable functional safety, localized service support, and has rich experience in mass production.
NeuSAR middleware and toolchain products build an open, flexible, and convenient development environment. The NeuSAR VMB message bus provides a unified interface for shielding the physical bus and communication protocol for different communication scenarios, making full use of the advantages of SOA as a service and supporting low-code and lightweight development of scenario functions.
For the development of innovative AI applications, NeuSAR upgrades the Python language development interface and WebService invocation mode to support the rapid import and deployment of cloud-native applications. At the same time, based on the powerful vehicle-cloud integration technology, it breaks the limitations of vehicle-end resources, realizes cloud resource sharing, and lays the foundation for AI general large language models to be used in vehicles.
At present, the NeuSAR software development platform has been mass-produced and applied in many car models, helping OEMs create innovative functional features and continuous iteration of functional experience, and has good adaptation and integration with many localized algorithms, communications, security, chips, microkernels and other enterprises to jointly provide car companies with software and hardware system solutions with lower cost and better quality.
During the summit, Infineon announced a new generation of the AURIX™ TC4x family, which addresses a wide range of automotive applications, including the strong need for functional integration in domain controller and zonal controller-based E/E architectures, providing enhanced connectivity, including advanced functional safety and security, to meet the needs of more high-level automotive applications and cross-application applications.
As the first software platform adapted to the TC4x series in China, Neusoft Reach NeuSAR made a grand appearance in the Infineon exhibition area, attracting extensive attention from government leaders, senior executives of car companies and industry partners.
As Infineon's "automotive electronics industrialization partner", Neusoft Reach and Infineon have carried out in-depth cooperation and mass production applications, continuously promoting the efficient application of innovative and mature software to complete vehicles, and realizing cross-platform and cross-model reuse. Based on the solutions provided by a variety of TC3 chips for different scenarios such as body, power, chassis, and autonomous driving domain controllers, NeuSAR has mass-produced nearly 30 products, covering nearly 40 models.
NewSAR cCore 4.0产品图
The combination of Neusoft Reach's NeuSAR and Infineon's TC4x in China, based on the Zonal core architecture and the central computing SoC, will be used for a wider range of domain control and central computing scenarios. NeuSAR cCore 4.0 is a new upgrade of Hypervisor, TSN, and ETHDriver software products, and TC4x to help reduce design risks and development costs.
In the future, Neusoft Reach will continue to uphold the ecological concept of sharing, openness, collaboration and innovation, work with the industry to build an open and standardized software architecture, interface specifications and application framework, create a prosperous and co-created automotive ecological picture, accelerate the AI model to bring new breakthroughs in product differentiation to the automotive market, and promote the automotive industry into a new future of more intelligent and sustainable development.