Developing Safety Case for Automated Driving following UL4600 & ISO5083

With the increasing in probabilistic outcomes by using complex algorithms/trainings for autonomous features, the safety of the product becomes more comprehensive where the safety case is one of claims to the overall safety assurance. Nonetheless, the safety case still needs to follow rigorous processes and provide all critical safety deliverables. Newer and updated standard such as UL4600 and ISO5083 have more considerations and claims can include in the safety case. From the definition of the autonomous feature to the risk acceptance criteria to the verification and validation of these features can all have elements from those new standards. ISO 5083 provides a framework for an autonomous system which can align with both ISO26262 and 21448. The scope and the fallback maneuvers can complement the hazards identified through safety analysis. The risk acceptance covers both Functional Safety and SOTIF through validation of potential metrics. Similarly, the evidence sufficiency and risk acceptance should be considered from UL4600. Major pieces of systems for an autonomous feature such as planning, sensing, perception, human interaction, etc. mentioned in UL4600 can provide additional reference in safety analysis for Functional Safety and SOTIF. Software tool certification and V&V coverage pointed out in UL4600 can become critical as more complex algorithms or software libraries are used for ML/AI. This paper will propose a safety case for automatic driving system using new information from UL4600 and ISO5083 under the frameworks of Functional Safety and SOTIF.