Enhancing SOTIF Analysis Using Model-Based Systems Engineering and Virtual Validation With Focus on Responsibility-Sensitive Safety

As automated driving systems become more widespread, ensuring their safety is paramount. The Safety of the Intended Functionality (SOTIF) analysis, specified in ISO 21448, plays a pivotal role in evaluating these systems' intended functionality, aiming to eliminate unreasonable risks arising from system insufficiencies. The complexity of contemporary autonomous systems necessitates a more systematic approach for SOTIF analysis. Furthermore, the Responsibility-Sensitive Safety (RSS) is an automotive safety framework, providing self-driving cars with decision-making capabilities akin to human drivers. However, despite RSS’ sophistication, potential failures due to real-world conditions and functional insufficiencies pose risks. This research proposes a Model-Based Systems Engineering (MBSE) methodology and a virtual validation strategy to enhance SOTIF analysis and advocates for integrating the SOTIF analysis during the RSS implementation to identify and mitigate risks. Using the Highway Pilot System with the integrated RSS framework as a case study, MBSE-SysML diagrams are employed to formulate specifications, perform safety analysis, and suggest functional modifications. An extensive verification and validation strategy focusing on RSS-relevant scenarios is developed using virtual validation techniques. This approach aims to improve the effectiveness of the SOTIF analysis process, ensuring the safety and the intended functionality of automated driving systems and also ensuring reliability while implementing the RSS logic in the autonomous vehicle. By contributing to the autonomous driving safety, this research lays the foundation for more robust SOTIF analysis methodologies with a focus on the RSS framework, enhancing the trustworthiness of autonomous vehicles.