For decades, the traditional systems design “V-cycle” worked well for automotive engineers. Based on a linear approach, the V-cycle assumes that all systems requirements and use cases are known upfront, and that the development process ends at sign off. But in the new era of connectivity (V2X, 5G) and increasing autonomy (powered by AI and deep learning), all scenarios and system interactions cannot be fully understood during the automotive engineering process. This reduces the V-cycle’s relevance and underscores the need for new ways of thinking about automotive engineering and design.
What’s needed is a new process for developing the connected AV (autonomous vehicle)—one where the system under development is not a single vehicle, but in fact an ever-changing ecosystem that can never be fully understood, because it’s always changing.
The Continuous Integration approach to the development and deployment of AVs offers a new spin on the traditional V-cycle. With this new approach, data generated and captured during a product’s development and operation becomes a valuable framework for the next system release. Like software design, the Continuous Integration process doesn’t necessarily end. Instead, it’s maintained over time.
Continuous Integration prepares engineers for a future we don’t yet fully understand. Like the old V-cycle model, this new approach leverages best practices and processes. But with Continuous Integration engineering, advanced design technology also plays a key role in the product lifecycle. Here are three critical technologies most essential for the era of Continuous Integration in AV development:
Comprehensive simulation models across the entire design lifecycle—The evolution of digital-twin technology has accelerated dramatically in recent years, and it now plays a central role in today’s model-based system engineering (MBSE) approaches. Powerful simulation software is now available to generate highly realistic, physics-based simulated raw sensor data for an unlimited number of potential driving scenarios, traffic situations, and other parameters—and this helps to dramatically reduce physical testing and speed verification and validation (V&V) cycles. Highly accurate simulated models can be quickly generated for virtually any onboard automotive system, including powertrains, tires, chassis, and electrical/electronic systems. And complementing all this sophisticated simulation technology is the availability of extremely comprehensive libraries of chassis systems and components, which facilitate the generation of limitless numbers of vehicle configurations. Finally, by leveraging a comprehensive suite of simulation technologies throughout the entire cycle, an ideal framework is established for model-in-the-loop (MiL), software-in-the-loop (SiL), hardware-in-the-loop (HiL) and full vehicle-in-the-loop (ViL) testing that is essential to successful V&V of first-generation AVs.
Real-world testing and data collection: it still matters—Despite the advancements in digital-twin technology, there will likely always be a role for real-world testing. Real-world data collection is essential to generating and maintaining scenario databases, which contain an “image” of everything that can happen on the road. These databases typically form the basis for the set of virtual scenarios that an automated driving system is designed to comprehend prior to its release. And of course, virtual test results can only be trustworthy if they are confirmed by physical testing. The key to success, however, is to keep time- and cost-intensive physical testing to a minimum. Having an established virtual and physical V&V framework is key in striking the right balance.
Robust requirements and data-management software—Designing and simulating the automotive systems of the future involves a vast array of technologies, processes, and teams working and collaborating around the clock to meet highly compressed deadlines. These factors, in turn, make it critical to properly select and deploy sophisticated, cloud-based product lifecycle management (PLM) software when developing first-generation AVs. A requirements- and data-management system should function as a backbone during the entire process. Leveraging a modern, adaptable PLM platform that securely connects people and processes across functional silos, with a digital thread for innovation, is the third essential technology for enabling Continuous Integration in AV development.
Verifying the performance of automated driving functions is particularly complex due to the infinite number of possible scenarios the car will encounter during its lifetime. Moreover, this set of potential scenarios will likely evolve continuously, since each new vehicle or device that hits the market will display its own unique behavior within the ecosystem. For this kind of dynamic application, a closed-loop V&V framework such as Continuous Integration can help ensure measurable test coverage and enable ongoing improvement.