Scholarlink

Hazards from complex operational scenarios bring huge challenges for autonomous driving (AD). This study proposes an External Operational Scenario-Systems Theoretic Process Analysis (EOS-STPA) approach, a novel safety analysis approach tailored for AD operational scenarios. Unlike traditional STPA method, EOS-STPA extends STPA to external operational scenarios. Moreover, by integrating ontology theory and employing a hierarchical control structure that encompasses closed-loop scenario control actions and feedbacks, EOS-STPA allows for formalizing the interaction between systems and operational scenarios. Furthermore, EOS-STPA identifies and generates formalized safety constraints comprehensively while enhancing Safety of the Intended Functionality (SOTIF) for AD. Additionally, EOS-STPA’s hierarchical control modeling facilitates efficient scenario hazard identification through structured scenario decomposition. As applied to an autonomous vehicle (AV) car-following scenario, EOS-STPA shows its capability in formalized safety analysis. This study marks the first extension of STPA to external operational scenarios while transforming technical system perspectives into operational scenario viewpoints.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Scenario Hazard Prevention for Autonomous Driving Based on Improved STPA

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Mingyang Zhao,
  • Ci Liang,
  • Tianxiao Wang,
  • Jinping Guan,
  • Long Wan

摘要

Hazards from complex operational scenarios bring huge challenges for autonomous driving (AD). This study proposes an External Operational Scenario-Systems Theoretic Process Analysis (EOS-STPA) approach, a novel safety analysis approach tailored for AD operational scenarios. Unlike traditional STPA method, EOS-STPA extends STPA to external operational scenarios. Moreover, by integrating ontology theory and employing a hierarchical control structure that encompasses closed-loop scenario control actions and feedbacks, EOS-STPA allows for formalizing the interaction between systems and operational scenarios. Furthermore, EOS-STPA identifies and generates formalized safety constraints comprehensively while enhancing Safety of the Intended Functionality (SOTIF) for AD. Additionally, EOS-STPA’s hierarchical control modeling facilitates efficient scenario hazard identification through structured scenario decomposition. As applied to an autonomous vehicle (AV) car-following scenario, EOS-STPA shows its capability in formalized safety analysis. This study marks the first extension of STPA to external operational scenarios while transforming technical system perspectives into operational scenario viewpoints.