Growing test volumes are not translating into greater confidence in vehicle safety behavior, according to test management consultant Maxim Zaretskiy, who argues that many OEMs have imported IT-style validation practices unsuited to safety-critical development. Across SIL, HIL and vehicle-level testing, regression suites, automation rates and coverage metrics keep rising, yet late-stage defects and release uncertainty persist. Zaretskiy points to patterns from large-scale integration programs where regression suites grow 30-40% within a single release cycle, while defect detection increasingly shifts to late vehicle-level validation.
The underlying issue, he suggests, is that coverage-driven validation—effective for IT systems where failures can be fixed post-release—does not map cleanly onto automotive constraints under ISO 26262, SOTIF and cybersecurity requirements. High traceability and broad regression coverage can still miss critical scenarios, particularly SOTIF-related cases where systems behave as specified but produce unsafe outcomes due to timing dependencies, sensor ambiguity or unforeseen combinations of valid states. Since not all hazards can be anticipated upfront, he argues for large-scale simulation and data-driven scenario discovery to surface long-tail risks.
Because safety-linked test cases cannot be removed without undermining traceability and safety cases, existing regression suites largely have to stay in place. The proposed shift is therefore not in test volume but in prioritization logic: moving from coverage metrics toward risk relevance, and from treating testing as pure execution toward using it to build system understanding of which scenarios actually define vehicle behavior.
Source: Automotive Testing Technology International — read the original article