Hazard analysis of an Automated Lane Keeping System using Systems-Theoretic Process Analysis
用系统理论过程分析方法分析自动车道保持系统，找出87个不安全控制行为和537个因果因素，提出1074条安全要求，对监管者和政策制定者尤其有用。
Systems-Theoretic Process Analysis (STPA) is an effective safety analysis technique that identifies how unsafe interactions among components within a complex system may result in accidents. This study aimed to evaluate the efficacy of STPA by applying it to an Automated Lane Keeping System (ALKS). The goal was to explore areas of potential risk in the system and make recommendations on how overall system safety could be improved. The STPA analysis of ALKS identified 87 Unsafe Control Actions (UCAs) based on interactions between the various components. An analysis of the UCAs revealed 537 causal factors (CFs), including software faults like flawed control algorithms and conflicting controls, sensor performance limitations, specification issues such as missing feedback signals, and errors in human–machine interaction, such as excessive dependence on the ALKS and drivers having incorrect expectations regarding ALKS operation. 1074 requirements were proposed to prevent or mitigate these causal factors, such as educating drivers about both the benefits and limitations of the ALKS to ensure safe use. The results highlighted the importance of communicating both the capabilities as well as the limitations of modern complex systems to the users to guarantee safety. This study, which is the first comprehensive application of STPA to ALKS, identified gaps with existing regulatory requirements for ALKS, and 87 recommendations were made to bridge these gaps. Our research has shown that this top-down, well-structured, and holistic method can especially be advantageous for regulators and policymakers to formulate requirements and policies to deploy and operate complex, innovative technologies, safely. • STPA identified performance limitations and human-system interaction errors. • Limitations of automated systems must be communicated to users. • Formalisation of STPA would ensure better coverage of system flaws. • UCA identification and quantitative assessment approaches have been proposed. • STPA can be used by regulators and policymakers for certification and policymaking.