Introduction: Understanding Software Transformation
When building software-defined vehicle (SDV) technologies, every decision influences delivery speed, system performance, and software longevity. Engineering choices are frequently presented as a compromise between speed and craftsmanship. While speed promises agility, faster time to market, and responsiveness to customer demands, craftsmanship and precision represent a tradition of automotive excellence that prioritizes reliability, safety, and refined driving experiences. From choosing between microcontrollers and high-performance computers to adopting service-oriented architectures, engineering choices frequently introduce friction. These decisions can slow progress and force teams into a false choice between moving fast and building with precision.
Can the industry accelerate SDV programs without compromising the safety, quality, and longevity that have always been its trademark?
This white paper explores:
- How an SDV development environment can be set up and supported to achieve both speed and craftsmanship
- The role of virtualization and the importance of shifting left in the development process
- The value of flexibility and reuse in accelerating innovation and reducing redundancy
- The need for alignment and automation to ensure consistency and scalability across teams and systems
Cycle linking customer needs, tech innovation, and business opportunities in software‑defined vehicles.
In the pursuit of speed, one of the most powerful strategies available to SDV engineers is virtualization: the ability to simulate, test, and iterate digitally before transitioning to physical prototypes. Virtualization, or simulation, is the digital adjacency that allows geographically and organizationally dispersed teams to collaborate seamlessly. By creating virtual representations of scenarios, vehicles, components, and processors, engineers can evaluate the impact of design choices across domains without waiting for physical hardware or going through the time-consuming process of manually assembling and testing every component. When combined with a shift-left mindset where testing, validation, and decision-making are moved earlier in the development process, virtualization enables earlier validation, faster feedback loops, and more informed decision-making across distributed teams.
When combined with a shift-left mindset where testing, validation, and decision-making are moved earlier in the development process, virtualization enables earlier validation, faster feedback loops, and more informed decision-making across distributed teams.
Virtualization spans multiple layers of the development process. Simulink® supports this approach across four key domains:
- Virtualization of scenarios
- Virtualization of the full vehicle
- Virtualization of components
- Virtualization of ECUs
Model‑based Design framework highlighting virtualization and simulation.
Virtualization of Scenarios
Engineering teams developing advanced driver-assistance systems (ADAS) and autonomous features must validate performance across a wide range of driving conditions. Physical testing, while essential, is expensive and inherently limited in scope. It cannot easily scale to cover the diversity of environments, edge cases, and traffic interactions required for robust validation.
At the same time, developing realistic scenarios requires significant effort. Tools like RoadRunner address this challenge by easing the work to design, visualize, and simulate complex driving scenarios. Scenarios from RoadRunner are compliant with OpenScenario and can be integrated into many popular third-party simulators, such as CARLA, CarMaker, dSPACE, and Unreal Engine.
Virtualization of the Full Vehicle
Validating embedded software against vehicle-level behavior is a critical step in SDV development. Here as well, virtualization allows you to simulate full vehicle systems under a wide range of conditions, long before prototypes are built.
MathWorks enables teams to build vehicle-level simulation models, such as STLASim from Stellantis, by providing products that support multidisciplinary design and analysis. These platforms allow teams to analyze performance across domains and can be used for a range of use cases, from system simulation, software development, and validation to system integration. You can zoom in on specific subsystems or zoom out to assess overall vehicle behavior using models that reflect varying levels of detail depending on the task.
Importantly, these simulation environments are designed for usability across diverse engineering roles with well-defined, user-friendly HMI, data analysis tools, and version control. Whether working in controls, mechanical systems, or software integration, teams can rely on a shared platform that supports their specific analysis needs.
This common foundation delivers multiple benefits, including:
- A reduced need for physical tests
- Lower development costs
- Improved quality due to earlier and more frequent validation
Virtualization of Components
Virtualization in SDV development is most effective when applied consistently across levels, from full vehicle systems down to individual components. Component-level simulation allows engineering teams to evaluate subsystem behavior in isolation while still maintaining alignment with broader system models.
For example, Simscape® extends this virtualization to energy storage systems. It offers parameterized models and design tools for battery management systems and can be directly reused for software validation.
Reuse of component models accelerates development and reinforces the goal of virtualization: enabling earlier testing, reducing reliance on physical prototypes, and improving confidence in system-level integration.
Sometimes, component models are developed by different teams using different tools. Rebuilding them for each workflow introduces delays and inconsistencies. Standards such as FMI and API-based integrations support interoperability across platforms, allowing validated models to be reused without modification.
Reuse of component models accelerates development and reinforces the goal of virtualization: enabling earlier testing, reducing reliance on physical prototypes, and improving confidence in system-level integration. We discuss model reuse in greater detail in the sections that follow.
Virtualization of ECUs
Virtualization of the ECU enables the decoupling of hardware and software development, allowing software teams to develop and validate target-ready software without being constrained by hardware availability. To address different development and validation needs, the industry has defined multiple levels of ECU virtualization ranging from L0, which focuses on application-level software validation, to L5, where the complete processor is fully virtualized. These levels and their associated use cases are defined in detail in the prostep ivip Association’s reference paper on virtual electronic control units.
At MathWorks, we support L0 and L1 levels of ECU virtualization directly within Simulink, enabling early-stage software development and validation. For higher levels of virtualization, MathWorks integrates with leading third-party platforms such as Vector, dSPACE, and Synopsys, allowing customers to scale their workflows based on system complexity and validation requirements.
MATLAB and Simulink enable automotive engineering organizations to accelerate vehicle development processes and to deliver vehicles that meet market requirements for safety, comfort, fuel economy, and performance.
In addition, MathWorks solutions can integrate with virtual processors, such as those provided by the Snapdragon Ride™ platform, making it possible to run full system-level simulations that combine scenario visualization, algorithm execution, and HMI rendering within a unified virtual environment.
Virtualization enables earlier testing and faster iteration, but to fully support speed and precision in SDV development, engineering workflows must also be adaptable. Flexibility and software reuse are critical, not just for efficiency but also for preserving validated work and accelerating delivery. By reusing proven components instead of rebuilding them, organizations reduce risk, maintain consistency, and ensure that progress scales as requirements change.
By reusing proven components instead of rebuilding them, organizations reduce risk, maintain consistency, and ensure that progress scales as requirements change.
This section explores key enablers of flexibility and reuse in SDV development:
- Flexibility across software, targets, and architectures
- Cloud-based collaboration and deployment
Flexibility Across Targets and Architectures
Service-oriented architecture (SOA) has emerged as a foundational paradigm for achieving over-the-air updates that characterize the software-defined vehicle. By structuring applications as collections of independent, self-contained units that are loosely coupled services, SOA enables teams to update individual components without disrupting the entire system. This modularity is particularly valuable in the context of software-defined vehicles, where frequent updates, over-the-air enhancements, and rapid feature deployment are now industry expectations.
Simulink provides a comprehensive workflow for developing service-oriented applications, enabling you to:
- Specify requirements
- Author architecture
- Model services
- Import existing components
- Reuse and generate production C++ code
This workflow supports deployment to a wide range of platforms, such as CPU, GPU, microcontroller, or microprocessor/SoC, or middleware, such as AUTOSAR Classic, Adaptive, and Proprietary, ensuring that services are reusable, upgradable, and scalable.
Cloud-Based Collaboration and Deployment
Distributed development teams need environments that support secure, scalable collaboration. These environments support controlled collaboration through secure interfaces, making it easier to share results, align on system behavior, and maintain development velocity across organizational boundaries.
Technology partners such as MathWorks help teams leverage the cloud by providing online tools, supporting connections to cloud-based data, enabling parallel simulations, and offering dashboards to monitor tasks running in the cloud. Cloud-based workflows also simplify deployment and scaling, helping teams respond quickly to changing requirements.
Perhaps the hardest challenge to overcome in SDV development is coordination across engineering disciplines. Software and systems engineers often operate with different assumptions, workflows, and priorities. Without alignment, these differences can slow progress and introduce inconsistencies. Also, while both software and system engineers value automation, automation without alignment would worsen the disconnect between these two disciplines.
This section explores three ways engineering teams can improve alignment and automate key workflows to support faster, more reliable SDV development:
- Using a shared language for architecture
- Simulation in CI pipelines
- Platform-level alignment
Model-Based Design connects systems engineering and software engineering mindsets through unified requirements, simulation capabilities, and automated development workflows.
A Shared Language for Architecture
When teams speak the same architectural language, they can bridge disciplinary gaps, streamline workflows, and deliver more reliable SDV systems faster.
One of the most effective ways to align diverse engineering teams in SDV development is through a shared architectural language. Architectural clarity ensures that everyone (systems engineers, software developers, and integration teams) works from the same understanding of system behavior, interfaces, and requirements. Without this common foundation, models often remain siloed, serving only as upstream documentation rather than guiding implementation.
System Composer™ addresses this challenge by providing a unified modeling environment for architectural design, analysis, and simulation. It enables systems engineers to define components, interfaces, and requirements in a structured format that software teams can interpret and act on. By connecting these models to simulation and code generation workflows, architecture becomes an active driver of development—not a static artifact.
This shared language preserves design intent across the lifecycle, reducing misalignment and accelerating collaboration. When teams speak the same architectural language, they can bridge disciplinary gaps, streamline workflows, and deliver more reliable SDV systems faster.
Simulation in CI Pipelines
In SDV programs, where software changes frequently and system behavior is complex, integrating simulation into continuous integration (CI) pipelines provides a practical way to validate functionality at scale.
Rather than relying solely on unit tests or static analysis, simulation-based CI workflows allow teams to evaluate software commits against realistic system models and driving scenarios. This approach ensures that changes are tested not just for correctness but also for their impact on overall vehicle behavior.
By embedding simulation into CI pipelines, teams create a feedback loop that supports rapid iteration while preserving engineering rigor. It enables early detection of integration issues, reduces manual testing effort, and helps maintain alignment between software and system-level requirements, all of which contribute directly to faster, more precise development.
Platform-Level Alignment
Bringing diverse engineering teams together requires more than shared models—it demands a common operational foundation. Platform-level alignment ensures that the tools, processes, and infrastructure supporting SDV development work consistently across disciplines. When automation scales beyond individual workflows to span entire platforms, it delivers its full value: enabling collaboration without friction.
Centralized platform engineering teams play a critical role in this alignment. By managing development pipelines, simulation environments, and integration workflows, these teams create a unified backbone that serves both systems and software engineers. This standardization reduces variability in how features are tested, validated, and deployed, allowing teams to focus on innovation rather than troubleshooting inconsistencies.
Tools that integrate simulation, code generation, and testing across domains complete the picture. They provide shared visibility into system behavior, minimize duplication of effort, and surface integration issues early, before they impact delivery. Coordinating workflows at the platform level not only preserves development velocity but also ensures quality and traceability throughout the lifecycle. In short, platform alignment transforms isolated efforts into a cohesive, collaborative process, bridging mindsets and accelerating SDV development.
As SDV development grows in complexity, engineering teams are looking for ways to accelerate workflows without compromising quality. Artificial intelligence (AI) is increasingly part of that solution: supporting faster decision-making, automating routine tasks, and enhancing the precision of both design and validation processes.
AI has been increasingly integrated into the development phase, with well-established use cases such as reinforcement learning for control design, virtual sensors that augment or replace physical sensors to improve reliability and reduce system cost, and reduced-order models that balance model fidelity with computational efficiency to enable faster simulations.
By embedding intelligence directly into the development process, [these copilots] support faster iteration and more informed decision-making, reinforcing the dual goals of speed and precision in SDV development.
Beyond development, AI has also established a strong presence in operations. Representative use cases include fleet data analytics for predictive maintenance and fleet management, anomaly detection in manufacturing lines to improve quality and uptime, and AI-driven feedback systems that close the loop between operational insights, design optimization, and continuous product improvement.
The next phase of AI adoption will be led by generative AI, enabling new ways to accelerate development workflows and improve overall engineering productivity. For Model-based Design, Simulink Copilot will be able to explain models and advise developers to correct errors. It will also support creation and automation of models as well as automation of several development steps.
Simulink Copilot provides generative AI–powered capabilities focused on Simulink and Model-Based Design. You can use Simulink Copilot to explain models and errors, learn tools and techniques, get design guidance, and automate predefined tasks.
Together, these copilots represent a shift in how engineering teams interact with their tools. By embedding intelligence directly into the development process, they support faster iteration and more informed decision-making, reinforcing the dual goals of speed and precision in SDV development.
The transformation to SDVs challenges us to rethink how we design, develop, and deliver value in an industry rooted in mechanical excellence. As engineering teams navigate this transformation, they face a series of decisions that often appear to demand compromise: speed or precision, agility or reliability, legacy or change.
This white paper has shown that these trade-offs are not inevitable. By aligning hardware and software development, automotive organizations can accelerate SDV programs without compromising safety or craftsmanship. Virtualization and shift-left methodologies enable early integration and validation, while service-oriented architectures and standards empower flexibility and reuse across diverse platforms. The synthesis of systems and software engineering—supported by automation and unified development platforms—bridges organizational divides and fosters a culture of collaboration and optimization.
MATLAB® and Simulink provide the means to create workflows that align hardware and software development, enabling engineering teams to iterate rapidly, validate comprehensively, and deploy innovative features with confidence.
The path forward is not about choosing between speed and craftsmanship. It’s about designing processes and selecting technologies that support both.
