Scholarlink

The paper addresses the problem of developing requirements and architecture for systems of modern highly automated vehicles (HAV) with a focus on the safety of such systems. The authors propose an AI-assisted algorithm (using LLMs and ontologies) as a methodological foundation for developers of AI assistants, enabling them to create tools for efficient safety tactic selection. The algorithm’s practical significance is empowering developers to build AI assistants. These, in turn, support architects in selecting tactics, making informed design trade-offs aligned with system requirements, and achieving benefits like proactive hazard mitigation and robust safety cases.

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

The Use of LLM in Selecting the Architectural Patterns for Safety Critical Automotive Systems

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Maxim Tikhomirov,
  • Oleg Kirovskii,
  • Gennadii Kruglov,
  • Alexander Luchkov,
  • Anton Korolev

摘要

The paper addresses the problem of developing requirements and architecture for systems of modern highly automated vehicles (HAV) with a focus on the safety of such systems. The authors propose an AI-assisted algorithm (using LLMs and ontologies) as a methodological foundation for developers of AI assistants, enabling them to create tools for efficient safety tactic selection. The algorithm’s practical significance is empowering developers to build AI assistants. These, in turn, support architects in selecting tactics, making informed design trade-offs aligned with system requirements, and achieving benefits like proactive hazard mitigation and robust safety cases.