Resilience engineering aims to embed resilience into socio-technical systems by addressing unforeseen situations and ensuring that systems can operate effectively under both expected and unexpected conditions. To this end, we propose an automatic monitoring approach for continuous usability evaluation, designed to detect and alert users to potential risks arising from user-system interactions. In our study, a proxy is integrated into an academic simulator of an onboard automated car system to monitor user actions through the generation of a Petri net-based activity trace. Usability evaluation is then carried out by comparing this trace with a real-time deduced task model in order to identify deviations from expected interaction patterns. Furthermore, machine learning techniques are employed to analyse the current situational context and to determine which tasks are permissible and which should be restricted. The overarching objective is to enhance system resilience by proactively monitoring user deviations and anticipating unexpected situations.
Jarraya et al. (Thu,) studied this question.