在自动驾驶安全案例中提供证据的形式化方法途径
A Formal-Methods Approach to Provide Evidence in Automated-Driving Safety Cases
AI总结:
本文提出用形式化方法为自动驾驶安全案例提供数学证据,证明功能安全需求向反馈回路组件的正确分解,并以可证明的形式化模型作为组件需求,减少评审、测试及数据与验证成本。
摘要:
自动驾驶系统的安全性必须通过有说服力的论证来证明,并由令人信服的证据予以支持,以说服认证机构、监管实体和公众允许这些系统在公共道路上行驶。这种说服通常通过将论证和令人信服的证据汇编成安全案例来促成。评审和测试是确保汽车系统正确性的两种常用方法,但无法探索通常无限的可能行为集合。相比之下,形式化方法是穷尽性方法,能够提供模型正确性的数学证明,并可用于证明系统组件的形式化模型满足功能安全需求的形式化表述。本文展示了形式化方法如何为功能安全需求到反馈回路中各组件的正确分解提供证据,以及该证据如何纳入安全案例的论证之中。如果获得证明,形式化模型将被用作对组件的需求。这种安全论证结构可用于减轻通过评审和测试来确保分解正确的需要,从而节省数据收集和验证时间方面的工作。
英文摘要:
The safety of automated driving systems must be justified by convincing arguments and supported by compelling evidence to persuade certification agencies, regulatory entities, and the general public to allow the systems on public roads. This persuasion is typically facilitated by compiling the arguments and the compelling evidence into a safety case. Reviews and testing, two common approaches to ensure correctness of automotive systems cannot explore the typically infinite set of possible behaviours. In contrast, formal methods are exhaustive methods that can provide mathematical proofs of correctness of models, and they can be used to prove that formalizations of functional safety requirements are fulfilled by formal models of system components. This paper shows how formal methods can provide evidence for the correct break-down of the functional safety requirements onto the components that are part of feedback loops, and how this evidence fits into the argument of the safety case. If a proof is obtained, the formal models are used as requirements on the components. This structure of the safety argumentation can be used to alleviate the need for reviews and tests to ensure that the break-down is correct, thereby saving effort both in data collection and verification time.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。