GenAI in Motion - Redefining the Automotive Product Development Landscape

The automotive industry is undergoing a period of unprecedented transformation. Electrification, autonomous driving, and software-defined vehicle technologies are reshaping the landscape, placing a premium on innovation and efficiency. Artificial intelligence (AI) and especially generative AI (GenAI) has emerged as a powerful tool for car manufacturers and suppliers, offering significant economic benefits and streamlining product development processes. This article explores the transformative impact of AI on automotive product development, introducing our comprehensive use case model and a practical approach to identifying and implementing the most valuable applications for your business.

Significant economic impact of GenAI on automotive product development

A recent McKinsey report called “The economic potential of generative AI - The next productivity frontier” tried to quantify the impact of GenAI across 63 use cases. I don’t want to go deeper into the “$2.6 to 4.4 trillion” that GenAI could add to the global economy. But I do like the finding that 75% of the value comes from use cases in the four areas of Customer Operations, Marketing & Sales, Software Engineering and Product R&D.

Figure 1: Impact of GenAI (Source: McKinsey)

McKinsey attributes the economic impact mainly to the potential of automating more than 50% of knowledge worker’s tasks. I would add the impact on time-to-market and on innovation: improved requirements and test cases as a rather mature GenAI use case can limit the number of changes, thus speeding up the development. This not only frees up capacity for innovation, AI can also support the generation and evaluation of new design ideas.

AI enables Data-Driven Engineering

Automotive is a data-rich industry: connected cars and smart factories produce immense amounts of data. Data-driven engineering is about use cases for this data in product development. At NTT DATA, we have developed a comprehensive model of 25+ AI use cases for product development, organized along the systems engineering V model and the phases of the product lifecycle. We describe each use case in terms of business functionality, required data and technological approach, for example GenAI, machine learning, natural language processing etc.

Figure 2: AI Use Cases in Engineering (excerpt) (Source: NTT DATA)

As an example, let’s take the use case of Patent Management. GenAI can support various processes such as patent search, classification, patent landscape analysis, summarization, drafting assistance and translation. This requires access to internal and external patent information. The technological approach would include large language models, natural language processing methods such as concept mapping, topic modeling and semantic search.

Value-Based Assessment of Use Cases

With limited resources and time, it is essential to identify and implement the high-value use cases for your organization. Our use case model uses three criteria for a first approach:

• Efficiency potential: The impact on efficiency within the automotive product development process. Considers factors such as time savings, cost reduction, resource optimization, and overall productivity improvements.
• Feasibility: Examines whether a use case is practical and achievable within the current technological and organizational constraints. Factors to consider include data availability, technology readiness, regulatory compliance, and integration challenges.
• Effort: evaluates the level of resources, including time, expertise, and investment, required to implement and maintain the use case successfully.

Figure 3: Value-Based Assessment of AI Use Cases in Engineering (Source: NTT DATA)

This kind of assessment allows us to concentrate on the upper-right corner, which represents use cases with high feasibility and significant efficiency potential. Examples in this quadrant include GenAI-based support for requirements engineering, as well as related test plans and test cases. Additionally, we find the well-established field of software development assistants, exemplified by Microsoft’s GitHub Copilot.

Conclusion

The digital transformation of the automotive industry is ongoing, and AI is a key tool for competitive advantage. NTT DATA is dedicated to collaborating with clients on this journey, providing expertise and support to navigate the opportunities presented by AI in automotive product development.

As a first step, we suggest our Proof-of-Value offering, which provides a hands-on experience with your priority use cases using your data within a pre-configured cloud environment. Feel free to contact us for further details on our AI solutions for engineering.

Note: this is a cross-post from the article on the NTT DATA website under GenAI in Motion - Redefining the Automotive Product Development Landscape | NTT DATA Group

Business Transformation for Software-Defined Vehicles

 What exactly is the transformative nature of software-defined vehicles (SDV)? This article explores transformation from a product development-perspective in three layers:

• automotive business including business models of various traditional and new players
• product from a "simple car" to smart connected SDV
• engineering processes, methods and tools for SDV development

Note: the article is based on my presentation at the "2023 Tokyo Systems Engineering Summit".

What is SDV?

There are many definitions of SDV. I will work with the following from the Eclipse SDV Working Group: "The term software-defined vehicle refers to a transformation where the physical and digital components of an automobile are decoupled (HW / SW separation) and features, functionality, and operations are defined through software. In a fully programmable car, digital components—such as modules for safety, comfort and infotainment, and vehicle performance—would be regularly developed and deployed through over-the-air updates."

The Business Case for SDV

For OEMs, the business case is pretty clear: SDVs enable additional revenue over the lifetime from a large fleet of connected vehicles by selling SW-based services and subscriptions, features on demand or data as a service. With large fleets such as in Volkswagen Group or Stellantis, this leads to expectations of 10-20 Billion Euro additional SW-enabled revenue annually by the end of this decade per OEM.

Business transformation in a system-of-systems

The value proposition of SDVs is a seamless integration into the digital lifestyle of customers. Connected cars have established a lot of basic components for this with a focus on connectivity to a proprietary OEM backend. The software-based services of an SDV require even more integration with non-automotive industries such as banking, insurance, utilities, telco and of course tech companies.

From a product development perspective, this demands systems engineering with a system-of-systems approach. The system-of-interest is no longer only the car, but the mobility system including services from providers in the system context. The system boundary and the interfaces to the service providers need to be actively developed.

Choosing the right standards – and increasingly open source initiatives – is a key to decouple the SDV from changes in the context. The right use of standards also increases the monetizable fleet across car lines and model generations.

Automotive companies need to build new business capabilities to develop and operate SDV. Please find more information on our EAM (Enterprise Architecture Management)-based approach including our NTT DATA business capability model in this article on Business Transformation Management.

SDV Product Transformation: Shift-North

In traditional E/E architectures, each function had it's own ECU. Developing functions on 100+ distributed ECUs becomes prohibitively complex and expensive. This led to E/E architectures with fewer domain controllers and/or zonal controllers on high-performance computers (HPC), running multiple functions in parallel on one device. Standardized operating systems and middleware provide the hardware abstraction that enables this shift.

SDVs continue this "shift-north" to upper layers of the stack, i.e. implementing functions on edge or cloud computing resources instead of incar devices.

This is the critical SDV product transformation because it allows the development of software independent from the hardware, resulting in:

• delivery of new SW-based functions and frequent updates over the air (OTA)
• very high scalability for resource-hungry applications such as AI and entertainment
• seamless digital user experience across channels and interfaces such as HMI, voice, web, app

Engineering Transformation: Shift-Left

Shifting activities to earlier phases of the engineering process, where change is faster and cheaper, is no new concept. Replacing hardware prototypes with simulation models has been a key concept of virtual product development for many years including the evolution of model-based systems engineering. SDV however continues this evolution with cloud-based virtual engineering workbenches, providing simulations of ECUs with binary parity and of the HMI, resulting in an improved developer experience for globally distributed teams.

Platforms and modular architectures have also been used as a "best practice" in product development for quite some time. For SDV, the physical platform of mechanical and E/E hardware is connected to a software platform in the cloud / backend. The business case of SDV, i.e. the size of the monetizable fleet, depends mainly on the number of SW-based services sold to as many cars as possible. Ideally, all brands, product lines and generations of cars are served by one software platform. This is enabled by another "shift-left", the investment into architecture & platform development.

The V model for systems engineering has been another "best practice" in product development for quite some time. The SDV adds agile BizDevOps concepts to the traditional V model:

The main transformation here is the extension of the V model from product development to the operations phase – hence BizDevOps. Software needs to be maintained over the lifetime. This is not only required by regulations such as UNECE R155 / R156 for software-update and cyber-security management. It is also the basis of the SDV business model of selling software-based services after the initial purchase of the car. Monitoring of the fleet including collection of monetizable data and management of cyber-security threats is a required capability for the SDV business model.

The cloud-based software functions are typically not as safety-critical and real-time-sensitive as the embedded ECU functions e.g. in body or powertrain domains. For speed and efficiency, these cloud-based software functions do not require the complete rigor of the V model and can be developed according to modern, agile software engineering practices. In true DevOps fashion, the ALM environment (Application Lifecycle Management) provides continuous integration and deployment (CI/CD), so that changes can be tested quickly with test automation and deployed into the fleet with OTA.

Conclusion: Truly a Transformation

SDV transforms the automotive business model, the actual product and the engineering processes, methods and tools for SDV development. It promises substantial software-enabled revenues, but also requires massive investments into new business and technology capabilities.

NTT DATA offers comprehensive support for the SDV transformation in the automotive industry:

Automotive Consulting

• Business Strategy & Transformation Consulting
• Systems Engineering
• Production & Aftersales
• Cybersecurity

Technology Services

• SW Development (Embedded & Cloud-native; Apps & Backend)
• Infrastructure (Hybrid Cloud / Data Center; 5G)

Please find more information under Industry page Automotive.

Note: this is a cross-post from my blog article under link