Vehicle SOA Tech Stack

The architecture of a modern tech stack for software-defined vehicles (SDVs) builds upon the principles of service-oriented architecture (SOA), carefully dividing the environment into safety-critical and non-safety-critical layers. The attached image illustrates the integration of cloud and on-board environments, categorized into Quality Management (QM), ASIL A/B, and ASIL C/D layers.

Layers of the Tech Stack for Vehicle SOAs

At the top of the stack lies the cloud environment, where middleware and applications are executed within a Platform-as-a-Service (PaaS) framework. The underlying layers of the cloud leverage Infrastructure-as-a-Service (IaaS) platforms powered by high-performance CPUs and GPUs. This configuration enables scalable computation and centralized management of services that interact with the vehicle's on-board systems.

Transitioning to the on-board QM environment, we see the use of container runtimes for efficient execution of microservices. These run within virtual operating system instances managed by hypervisors, which typically utilize general-purpose OSs such as Linux. This layer relies on high-performance compute environments powered by CPUs and GPUs, ensuring the rapid execution of non-critical vehicle functions and enhanced vehicle experiences.

For the ASIL A and B environments, the stack incorporates specialized platforms such as Autosar Adaptive or the Robot Operating System (ROS). These operate on generic microprocessors to support safety-critical functionalities while maintaining sufficient flexibility for dynamic service orchestration.

The ASIL C and D environments delve deeper into real-time operations, where the tech stack includes high-end ECUs running real-time operating systems on generic microprocessors for zone controllers. At the endpoints, the stack employs low-end ECUs powered by microcontrollers. These typically run Autosar Classic or similar micro-OS platforms, ensuring the safe and reliable execution of highly critical tasks like braking and engine control.

This layered and modular approach to the SDV tech stack highlights how safety-critical and non-critical environments can coexist. Each layer is optimized for its specific role, from supporting scalable cloud-based computation to enabling real-time, safety-critical operations on embedded systems. This separation ensures scalability, functional safety, and the agility needed for software-defined vehicle ecosystems.

Application Perspective on the SDV Tech Stack

To better understand the modern SDV tech stack, let’s revisit it from the perspective of applications. Previously, we introduced two distinct applications: one for a mobile mechanic and another for an on-board passenger welcome sequence. These applications demonstrate how the tech stack enables seamless integration between various components and ensures safety and functionality.

The first application, designed for a mobile mechanic, runs on a smartphone. This app allows the mechanic to open the vehicle door remotely, making use of the vehicle-to-cloud API to communicate with the car. The app sends a command to the cloud, where it interacts with a microservice implementing the door open API. This microservice, running in the cloud runtime, processes the request and transmits it to the vehicle's on-board system via the vehicle-to-cloud API.

The second application, the passenger welcome sequence, operates directly on the vehicle within the QM container runtime environment. This app is responsible for creating an engaging user experience by adjusting settings, such as seat position, and opening the door when the owner approaches the car. Like the mobile mechanic app, it leverages the door open API to perform its functions.

In both scenarios, the door open API acts as a central interface, abstracting the complexities of interacting with the vehicle’s hardware. The implementation of this API involves a multi-step process to ensure functional safety. When a request is made to open the door, the API must first pass through the signal-to-service API, which connects it to the embedded environment. In the embedded environment, safety checks are conducted in the following sequence:

1. Vehicle Speed Check: Ensures the car is stationary before proceeding.

2. Rear Traffic Monitoring: Uses the rear camera and AI to detect any approaching vehicles.

3. Side Clearance Check: Verifies that no obstacles or pedestrians are near the door using side cameras.

4. Execution at the ECU Level: After all safety checks are validated, the signal-to-service API communicates with the embedded ECU to physically unlock and open the door.

This layered approach showcases how the SDV tech stack supports end-to-end functionality, ensuring safety-critical computations are performed within appropriate environments while exposing reusable APIs for application development. The ability to reuse the door open API across both on-board and off-board applications highlights the modularity, scalability, and interoperability that a service-oriented architecture brings to software-defined vehicles.

Case Study: Rivian's Modular Architecture and Strategic Vision

To conclude, let’s revisit Rivian’s innovative approach to vehicle architecture and its broader implications. Rivian's vision revolves around modularity, where different versions of electrical hardware are abstracted through a hardware adaptation layer. This approach allows Rivian to build flexible, generic software layers on top, which can be tailored for specific vehicle variants.

This modular architecture is crucial not only for Rivian’s ability to scale across its product portfolio but also for its strategic collaboration with Volkswagen. In the context of the Rivian-Volkswagen joint venture, this architecture provides an opportunity for Volkswagen to leverage Rivian’s core platform while introducing distinct digital features tailored to its own brand and market requirements. For instance, Volkswagen could reuse Rivian's higher-level architecture for vehicle operations but customize it with proprietary digital experiences, enhancing its competitive differentiation in areas such as infotainment, user interaction, or advanced driver assistance systems.

This partnership highlights the transformative potential of scalable and flexible architectures in the automotive industry. By abstracting hardware complexities and focusing on software differentiation, Rivian’s approach sets a benchmark for the efficient and collaborative development of next-generation, software-defined vehicles.