Original title: Virtual Testing of Automated Driving Systems through Credible Simulations
Riccardo Dona, Espedito Rusciano, Biagio Ciuffo · September 2026
Testing self-driving systems only on real roads is impractical because their behavior is complex and real-world conditions are vast. Instead, regulators use virtual testing on computers to approve automated driving systems. However, officials must know if these computer models are trustworthy. Currently, teams rely only on validation, which checks if a simulation matches reality. This method does not scale well across complex setups that use multiple digital tools. To solve this problem, researchers developed a risk-based framework based on safety practices like NASA's STD-7009 standard. The framework links credibility requirements to how people use the results and how critical the decision is. It covers the full lifecycle, tracking tool management, assumptions, limits, verification, validation, and sensitivity analysis. The framework adjusts its rules based on whether a simulation explores ideas, aids a decision, or completely replaces real-world tests.
Imagine you're following a recipe to cook a meal for a casual dinner versus a high-stakes party. For the casual dinner, if the dish is slightly off, it's no big deal. But for the party, every ingredient and step must be precise because mistakes could ruin the event. In this paper, simulations for self-driving cars are like those recipes. The framework decides how strict the checks should be based on whether the simulation results influence critical safety decisions, ensuring higher scrutiny when the stakes are higher.
Regulators and car makers rely on simulations to approve self-driving systems, but they need a reliable way to ensure these simulations are accurate. This framework gives them a structured method to assess credibility based on how critical the safety decisions are, improving road transport safety.
The abstract doesn't state specific limitations. It mentions the framework is demonstrated for ADS and directly applicable to road safety studies where virtual testing is central.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Virtual Testing of Automated Driving Systems through Credible Simulations — Riccardo Dona, Espedito Rusciano, Biagio Ciuffo
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.