arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2209.14036cs.LOcs.FLcs.SE
收藏

基于时间自动机形式化建模与验证的数字公路法规

Towards a Digital Highway Code using Formal Modelling and Verification of Timed Automata

Gleifer Vaz Alves, Maike Schwammberger

摘要:

设计安全、可靠和值得信赖的自动驾驶汽车(AV)的挑战之一是确保AV遵守交通规则。为此,自动驾驶汽车需要能够理解和推理交通规则。在以前的工作中,我们引入了空间交通逻辑USL-TR,允许明确的,机器可读的,形式化的交通规则。这只是迈向可验证地遵守交通规则的自主交通代理的第一步。在本研究预览中,我们重点关注两个进一步的步骤:a)直接从交通规则中检索行为图,b)将行为图转换为使用USL-TR公式的时间自动机。有了这个,我们就有了一个正式的交通规则代表,并可以朝着建立数字公路代码的方向迈进。我们简要地设想进一步的步骤,其中包括添加环境和代理模型的时间自动机,最终实现和验证这些交通规则模型使用选择的形式化验证工具。

英文摘要:

One of the challenges in designing safe, reliable and trustworthy Autonomous Vehicles (AVs) is to ensure that the AVs abide by traffic rules. For this, the AVs need to be able to understand and reason about traffic rules. In previous work, we introduce the spatial traffic logic USL-TR to allow for the unambiguous, machine-readable, formalisation of traffic rules. This is only the first step towards autonomous traffic agents that verifiably follow traffic rules. In this research preview, we focus on two further steps: a) retrieving behaviour diagrams directly from traffic rules and b) converting the behaviour diagrams into timed automata that are using formulae of USL-TR in guards and invariants. With this, we have a formal representation for traffic rules and can move towards the establishment of a Digital Highway Code. We briefly envision further steps which include adding environment and agent models to the timed automata to finally implement and verify these traffic rule models using a selection of formal verification tools.

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。

↑