The development of autonomous vehicles presents the automotive industry with numerous challenges related to safety and functionality. Key issues include the definition of functions, safety evaluation of AI/ML systems, and addressing systematic faults, particularly in the context of safety concepts such as SOTIF. These concerns are heightened by gaps in addressing hazardous known and unknown scenarios during hardware and software verification, as well as risks related to tooling. Standards such as IEC 61508, ISO 26262, ISO 21448, ISO/TS 5083, and ISO PAS 8800 provide structured solutions to these challenges, but integrating their objectives into practical, real-world applications within limited timeframes remains a daunting task. This presentation will explore pragmatic techniques to address these challenges, including integrated risk analysis, the application of “safe by design” principles, and the adoption of best practices derived from commonalities across safety standards. Additionally, we will propose solutions tailored to AI/ML safety and functional safety issues in autonomous vehicle systems.

In this session, you will:

• Get a better understanding of safety gaps in autonomous vehicle development and identifying potential risks
• Gain insights into addressing functional safety challenges using standards like ISO/TS 5083
• Learn more about strategies for applying integrated risk analysis, V&V safety solutions, and tool chain-centric approaches
• Uncover approaches to overcoming AI/ML safety challenges in autonomous driving systems