AI-Assisted Functional Safety Verification and Validation under ISO 26262
Keywords:
ISO 26262, Functional Safety, Verification and Validation (V&V), Artificial Intelligence, Machine Learning, Automotive Cyber-Physical SystemsAbstract
The escalating complexity of automotive electronic systems, particularly with the integration of Advanced Driver Assistance Systems (ADAS) and Autonomous Driving (AD) capabilities, has pushed traditional functional safety frameworks to their limits. The ISO 26262 standard provides a structured lifecycle for ensuring functional safety, but its reliance on manual verification and validation (V&V) processes introduces significant bottlenecks, escalating development costs, and risks of human error. This paper presents a comprehensive, AI-assisted framework designed to optimize and automate V&V workflows across the ISO 26262 lifecycle. By leveraging Large Language Models (LLMs) for automated requirement traceability, machine learning classifiers for Failure Mode, Effects, and Diagnostic Analysis (FMEDA), and deep reinforcement learning for automated test case generation, the proposed methodology dramatically reduces manual overhead while improving fault injection coverage. A rigorous experimental evaluation was conducted on an industrial-grade electric vehicle powertrain control system. The results demonstrate that the AI-driven framework achieves a substantial reduction in requirement mapping time, an exceptional accuracy rate in automated failure mode classification, and a significant increase in critical edge-case fault coverage compared to conventional manual techniques. Furthermore, the framework provides full auditability and explainability, ensuring compliance with strict safety-critical automotive auditing standards. This research bridges the gap between advanced artificial intelligence and rigorous functional safety compliance, establishing a scalable paradigm for next-generation autonomous vehicle engineering.
References
1. Ahmad, M., & Khan, S. (2022). Model-based functional safety engineering for autonomous vehicles: A systematic literature review. Journal of Systems and Software, 186, 111195.
2. Amersbach, C., & Kloos, H. (2021). Functional safety of automotive systems: Principles and practices according to ISO 26262. Springer Embedded Systems Series. Springer.
3. Becker, J., & Nemati, F. (2023). Natural language processing for automated requirements traceability in safety-critical cyber-physical systems. IEEE Transactions on Software Engineering, 49(3), 1204–1218.
4. Chen, Y., & Zhang, L. (2024). Reinforcement learning-based automated test case generation for autonomous driving software. Automotive Innovation, 7(1), 45–58.
5. Devlin, J., Chang, M. W., Lee, K., & Toutanova, K. (2019). BERT: Pre-training of deep bidirectional transformers for language understanding. Proceedings of NAACL-HLT, 4171–4186.
6. Gupta, R., & Kumar, P. (2023). Automated safety analysis of automotive E/E systems using ensemble machine learning classifiers. Reliability Engineering & System Safety, 230, 108912.
7. International Organization for Standardization. (2018). Road vehicles — Functional safety — Parts 1 to 12 (ISO Standard No. 26262:2018).
8. Kim, H., & Lee, S. (2022). Deep learning-based failure mode and effects analysis for advanced driver assistance systems. International Journal of Automotive Technology, 23(4), 1015–1026.
9. Lundberg, S. M., & Lee, S. I. (2017). A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems, 30, 4765–4774.
10. Martinez, M., & Soler, A. (2023). Evaluating fault injection techniques in hardware-in-the-loop environments for ASIL D automotive compliance. IEEE Access, 11, 34567–34581.
11. Nguyen, T., & Dang, M. (2024). Overcoming data scarcity in safety-critical domains: An application of SMOTE-ENN for FMEDA automation. Journal of Safety Research, 89, 201–214.
12. Park, S., & Oh, C. (2023). Explainable artificial intelligence (XAI) in automotive functional safety auditing: Opportunities and hurdles. Safety Science, 165, 106190.
13. Siddiqui, A., & Harrison, R. (2022). Requirements engineering challenges in software-defined vehicles: An industrial case study. Requirements Engineering, 27(2), 243–259.
14. Wang, J., & Liu, X. (2025). Safety of the Intended Functionality (SOTIF) verification via adaptive reinforcement learning stress testing. IEEE Transactions on Intelligent Transportation Systems, 26(2), 1450–1464.
15. Zhao, K., & Zhou, Y. (2024). Continuous integration and automated compliance frameworks for ISO 26262 within agile development workflows. Software Quality Journal, 32(1), 89–115.