To deliver a highly autonomous vehicle, usually called automated driving system (ADS), there are different dimensions of requirements that need to be met to guarantee safety and acceptability. This concerns basically everything from legal aspects, standards, and customer expectations. As with many fields of application, new paradigms exist to develop and maintain software products. Just to mention a few: 1) Using AI in software development, e.g., for the perception of the environment, 2) Continuous Integration and Continuous Deployment (CI/CD) as an enabler for an efficient ODD expansion, i.e. frequently your car becomes better and can handle ever more demanding situations, and 3) the agile way of working, allowing for a more flexible approach with the opportunity for updates as increasing amounts of data is available from the fleet. These new paradigms are also bearing some challenges, for example, an extensive architecture that we need to deal with, as well as compliance with standards that are not always in line with the agile way of working, but also new standards that focus on AI. How should we be able to argue that all components and their integration is safe, or rather, that we have a predictive safety case?

The talk will give an overview of standards that needs to be considered for ADS and reflect on some of the challenges w.r.t. the new paradigms, specifically as inspiration to the requirements research community.

Dr. Håkan Sivencrona is Chief Safety Officer at Zenseact, a Volvo car subsidiary. Zenseact focuses on ADAS and (un)supervised driving.

Håkan has worked with ADAS and AD since 2005 and holds a PhD from Chalmers University from 2004. Currently, he is working with strategies for self-driving and is also involved in several research projects and standardizations, such as ongoing TS 5083 and UL 4600. In his standardization commitment, he is one of the founding fathers of the SOTIF-standard (safety of the intended function) and the SEooC part of ISO 26262, Safety Element out of Context.

Dr. Alessia Knauss is a requirements engineer and researcher at Zenseact, where she is involved in several research projects related to human factors, driver-autonomous vehicle interactions, and requirements engineering for AI-based systems.

Alessia received a PhD degree on the topic of Requirements Engineering for Adaptive systems from the University of Victoria, Canada in 2015. She has been a Postdoctoral fellow at Chalmers University of Technology, investigating testing challenges for autonomous vehicles before working with Autoliv/Veoneer as a research specialist and requirements engineer. Her main research interest lies at the intersection of users and autonomous systems/IT Ecosystems, including system adaptation to user needs with the help of AI.