Introducing ODD-SAF: An Operational Design Domain Safety Assurance Framework for Automated Driving Systems

Over the past years, there has been an increasing acceptance on the need for scenario-based testing to ensure safe performance of Automated Driving Systems (ADS). This is a departure from earlier conceptions where number of miles driven was being considered as the only way of demonstrating safety. As ADS show a great degree of variety in their complexity, use cases, as well as Operational Design Domain (ODD), a scalable and pragmatic approach for safety assurance of ADS, the ODD-SAF, is therefore proposed. The ODD-SAF relies on the ODD description and leverages on the EU and UNECE discussions around ADS safety requirements and assessment methods to generate behavioural competencies for the overall safety assurance. The approach extends with the identification of test scenarios, classified into nominal, critical and failure types, and pass-fail criteria, leveraging for example the concept of rules of the road and safety models for driving behaviour. It is suggested that a SAF should incorporate each of the categories of the test scenarios to ensure confidence in the performance of the ADS.