Navigating the software-defined vehicle revolution requires transitioning from hype to reality through continuous testing, caution and cross-functional collaboration
According to Gartner, new technology often goes through a hype cycle in which its initial potential creates great expectations. More often than not, the hype is followed by the so-called ‘disillusionment’, especially when the complexity of the technology extends the time required for development. Eventually, the various challenges are resolved, enabling commercial adoption. The SDV is currently riding the crest of its hype cycle. Companies like Tesla, Nio and Rivian have made progress due to their clean-sheet SDV architecture. In contrast, traditional OEMs are on an evolutionary journey, navigating the crests and troughs of the technology hype cycle.
One of the key challenges for them is that their product development process is linear, with fixed milestones. At the end of each milestone, the development team is dismantled and moves on to the next vehicle program.
In contrast, a software-centric product development approach operates on a paradigm of continuous integration/continuous deployment (CI/CD). An associated challenge with this is that conventional hardware-centric development is organized by discrete subsystems, which can be designed and tested somewhat separately before integration. Software-centric development, by definition, requires more up-front cross-functional participation, especially where the software is interdependent on many functions.
Transitioning from hardware-centric development to SDV development requires a shift from the traditional V-framework of fixed design-development-testing to a new framework of a continuous cycle, as practiced in consumer electronics and other software-dominant products. However, this CI/CD must be done without sacrificing vehicle safety. Unlike consumer devices that can occasionally fail without major impact, a vehicle failure can lead to a catastrophe, including loss of life.
This transition necessitates that product development teams embrace the concepts of ‘shift left’ (test early) and ‘shift right’ (test and monitor in production), ensuring that designers, developers and validators can work together in a more agile environment.
The need for continuous testing becomes more pronounced with the rollout of over-the-air updates. Recent incidents have shown that OTA updates intended to improve user interfaces can inadvertently cause failures in other systems, occasionally leading to the ‘bricking’ of infotainment systems. These incidents highlight the importance of fast, continuous in-house testing and controls in the software update process. Testing for SDV software modules must adhere to the principle of ‘do no harm’ – validating that one software component does not adversely affect another.
Changes in software may also create new vulnerabilities in performance and security. New test processes must catch such vulnerabilities at every stage of product development or updates, which is something that the auto industry, with its long and linear product development process, previously had sufficient time to ensure. Software engineers making even small changes must receive immediate feedback from the system, particularly regarding the likelihood of adverse side effects or harm to other vehicle systems.
The need for fast and continuous testing has ramifications, including the need for up-front simulation, using SIL/HIL test rigs that can run immediately to capture any unforeseen issues. Concepts that must be embraced to scale SDV deployment include the use of digital twins and shadow mode testing. By creating digital twins of key systems early, engineers can simulate, test and validate various scenarios without waiting for late-stage validation.
Some auto developers use ‘shadow testing’ or ‘shadow mode testing’ to fast-track software-based systems testing. This needs extra hardware resources so the system can run both the current and the updated software simultaneously. Outputs can be compared to detect potential issues in the new software, giving developers an early sense of the validity of the new version directly from the field.
The new generation of automotive leaders needs to combine the software and hardware expertise of the tech sector with a full appreciation of the safety and reliability required to build the next generation of vehicles. They must maintain the failsafe, well-validated reliability of today’s automobiles. Leadership and talent reshuffling within legacy OEMs must balance the infusion of software skills without sacrificing traditional safety-centric development capabilities.
Dr Partha Goswami, who is principal of PG Mobility Analysis, joins the ATTI Awards judging panel this year. He brings more 30 years of experience in the auto industry in R&D, planning and technology strategy.