For high-level autonomous vehicles, under many circumstances, accidents are not
caused by functional failures, but by system performance limitations and human
misuses. ISO 21448 defines this kind of safety issue as safety of the intended
functionality (SOTIF). However, most of SOTIF-related researches focus on hazard
identification and risk evaluation, as well as verification and validation.
There is still lack of public systematic identification method for triggering
events. Therefore, this paper proposes an identification framework for
triggering events which are related to system performance limitations and human
misuses. In the aspect of system performance limitations, there are four steps,
which are the specification of the research target, the analysis of influencing
factors, the identification of triggering events, and the verification test. In
the aspect of human misuses, triggering events can be identified based on the
guide words for human error. Furthermore, based on the identification of
triggering events, a safety analysis and verification framework for both
high-level and low-level autonomous vehicles is proposed. The proposed framework
contains five steps, which are hazard identification and risk evaluation, the
identification of triggering events, the generation of test cases, test and
results analysis, and the derivation of functional safety requirements. Besides,
a SAE L3 autonomous vehicle is analyzed and corresponding tests are conducted.
Two tests are illustrated in detail in this paper, which involve triggering
events related to system performance limitations and human misuses,
respectively. And the results show that the two triggering events could both
lead to effective hazardous events, which proves the feasibility of the proposed
framework.