By ATN
Author: Evangelos Simoudis, Managing Director at Synapse Partners.
Building The Software-Defined Vehicle: Core Technologies For Zonal Architectures And Intelligent User Experiences
In my book The Flagship Experience, I define a Software-Defined Vehicle (SDV) as a vehicle whose functions are primarily enabled through software. SDVs differ from conventional vehicles over five distinct dimensions. They enable continuous improvements of the vehicle and its in-cabin experience through over-the-air (OTA) software updates; support higher levels of driving automation; enhance the electric powertrain through intelligent battery management; enable the control, optimization, and management of a fleet; and facilitate the vehicle’s interaction with the transportation infrastructure. A prime example of an SDV capability is the self-parking function of Ford vehicles.
While early examples of software-defined vehicles have been developed using domain-based architectures, I assert that zonal architectures provide more benefits and, together with AI, enable a new automotive customer experience. A zonal architecture organizes the vehicle’s electrical and electronic (E/E) systems into a small number of independent zones. Each zone is controlled by a powerful zonal controller, which handles the software and hardware for that specific area of the vehicle. Rivian’s R1, NIO’s NT2, and XPeng’s G9 platforms are based on zonal architectures.
Zonal architectures incorporate fewer but more powerful controllers than domain-based architectures. The components of these architectures are designed from the ground up to be updatable over the air, resulting in a process that is simpler and faster to perform and manage. This means that overall, the vehicle can be easily updated and reconfigured at the lowest level, the architecture can scale across models, and the complexity of the vehicle’s wiring is reduced positively impacting manufacturing costs and vehicle weight. Because of their hardware and software architecture, these vehicles generate more easily accessible data.
The Software-Defined Vehicle relies on three types of software: the Software Platform that makes its operation possible, applications, such as ADAS, that enhance the vehicle’s basic functionality, and the Software Tooling Platform, which includes the tools used by all the processes before the vehicle being delivered to the market. The Software-Defined Vehicle and the customer experience the OEM wants to offer around it, which I define as the Flagship Experience, are tightly linked, such that all the software needs to be built as a comprehensive and well-integrated system that can be leveraged across all vehicle models and incrementally improved over time. To achieve its software goals, the automaker must establish consistency across all software (a “software philosophy”), develop a detailed architecture for the entire software system, and adopt an agile software development process for its implementation including DevOps practices and Continuous Integration / Continuous Delivery (CI/CD) tactics.
The need for a unifying consistency across the Software-Defined Vehicle and the customer experience means that the overall architecture governing the vehicle and experience must:
- Simultaneously encompass all the vehicle’s layers (skateboard, hardware platform, and software platform) and the services offered, and
- Unify data, software, hardware, APIs, and security.
The software architecture must not only enable configurability, extensibility, and upgradability but also inform the automaker’s other processes (manufacturing, support, servicing, etc.) that benefit from understanding the impact of every modification of the vehicle’s initial configuration because of fixing a problem, or adopting, removing, or upgrading a service.
The SDV’s Software Platform consists of middleware, a real-time operating system, the infotainment operating system, and APIs. This platform can be implemented using a service-oriented architecture (SOA), in which the desired functionality is provided by microservices. A services-oriented architecture requires that the Software-Defined Vehicle is equipped with high-performance CPUs and high-bandwidth Ethernet-based networks.
The development of the zonal SDV requires a variety of software tooling: a data platform, digital twins, intelligent simulators, a machine learning stack, a software and machine learning model version control system, and Integrated Development Environments (IDEs) for vehicle applications and services. The tooling may run on the OEM’s private cloud or on hybrid clouds to facilitate the collaboration of the automaker’s distributed teams, as well as the teams that are part of the automaker’s worldwide partner ecosystem.
AI plays a transformative role in every aspect of the software-defined vehicle. It enables autonomous driving, controls the battery management system thus impacting the vehicle’s range, and personalizes the in-cabin experience. For example, the infotainment system may include machine learning models for natural language understanding. The vehicle’s automated driving is enabled by a different set of machine learning models and so is the battery management system. A machine learning stack must include components for data ingestion, data engineering, data science, and operation (MLOps).
The development and deployment of the customer experience described in The Flagship Experience require a variety of AI software, including AI algorithms presented in the book. Some of this software is deployed in the cabin, such as the driver monitoring and cabin personalization software, while other is deployed in the cloud, such as software that constantly analyzes a vehicle’s operation to predict its maintenance needs or even software that analyzes and optimizes the operation of a vehicle fleet. The integration of zonal architecture and AI technologies within the SDV marks a significant leap forward for the automotive industry. As these innovations mature, vehicles will become better at providing an intelligent, driver- and passenger-focused experience. Engineers and technologists at the forefront of this shift have an unprecedented opportunity to design vehicles that align with evolving consumer expectations and safety standards.
