An Stpa-Based Formal Composition Framework For Trustworthy Autonomous Maritime Systems

SAFETY SCIENCE(2021)

引用 22|浏览16
暂无评分
摘要
A key risk with autonomous systems (AS) is the trustworthiness of the decision-making and control mechanisms that replace human control. To be trustworthy, systems need to remain safe while being resilient to unpredictable changes, functional/operational failures and cybersecurity threats. Rigorous validation and verification are essential to ensure trustworthiness of AS. Current engineering practice relies heavily on Verification and Validation (V&V) test-and-fix of system characteristics which is very time-consuming and expensive, limiting the possibilities for exploration of alternatives in system design. Instead, we focus on building a correct-by-construction system. In this paper, we present an approach to identifying and analysing mission requirements for squads of autonomous missions. Clear definition of requirements is an important pre-requisite for mission planning and for V&V of mission management. We use a structured approach to requirements identification and use formal modelling to help remove ambiguities in the requirements and to specify formal properties that should be satisfied by the missions. Our approach use a combination of analysis techniques based on Systems Theoretic Process Analysis (STPA) and formal modelling to generate critical requirements that ensure the safety and security of the system. We also suggest a compositional approach for formal modelling to enhance re-usability and address the complexity of formal modelling. Our approach is being evaluated through consideration of a combined mission of an Unmanned Surface Vehicle (USV) with deployment/recovery of small Unmanned Underwater Vehicles (UUV) within a shipping channel whereby the USV has to safely maintain station for a long period and then proceed to recover the UUV, while maintaining a communication link to an Unmanned Aerial Vehicle (UAV).
更多
查看译文
关键词
Formal methods, STPA, Event-B, Requirements, Maritime autonomous systems, SE-STPA
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要