Safety Systems Engineering in Autonomy

Safety systems engineering in autonomy is concerned with how engineers formally demonstrate that complex, software-intensive systems — autonomous vehicles, aircraft, medical devices — will behave safely even under failure or unexpected conditions. The work draws on structured methodologies such as ISO 26262 for automotive functional safety and Automotive SPICE for process assessment, alongside techniques like model-based development and assurance cases, which provide traceable arguments linking design decisions to safety claims. As autonomy introduces machine-learned components whose behavior cannot be fully specified in advance, the field is actively grappling with how to extend traditional certification frameworks — built around deterministic, hand-coded software — to cover statistical and adaptive systems. Open questions include how to construct rigorous assurance cases for neural networks, how to integrate cybersecurity threats into functional safety analyses, and how much evidence regulators should require before granting approval for systems that continue to learn after deployment.

Works

26,295

Total citations

68,206

Keywords

Safety AssuranceAssurance CasesSoftware CertificationFunctional SafetyAutomotive SPICEModel-Based Development