An Integrated Framework for the Assessment of Automated Driving Systems

The road transport system is undergoing major transformations, notably the transition to fossil-free propulsion and advances in automation. In the automotive domain, highly automated vehicles are typically referred to as incorporating Automated Driving Systems (ADS). Despite significant technological progress and anticipated benefits, the widespread deployment of these systems remains limited. This slow rollout is largely due to the increased complexity at higher automation levels, where vehicles must take on the full driving task and transition to fail-operational architectures capable of managing faults without human intervention. The increasing complexity of these systems stems from a growing reliance on new, often noisy sensors and external connectivity for perception and decision-making. In addition to established safety concerns, cybersecurity has become essential, as communication interfaces and complex sensors introduce new vulnerabilities. Advances in environmental sensing further require attention to functional limitations that arise independently of malicious interference. Taken together, these developments present challenges in forming a comprehensive understanding of the contexts in which the systems are intended to operate safely. This, in turn, hinders the development of effective validation procedures and complicates efforts to substantiate claims of safety, security, and overall trustworthiness. This PhD thesis advances the assurance of ADSs by integrating safety and cybersecurity into a unified assessment framework. The contributions are threefold: (1) context-dependent requirements; introducing technology-specific assessment templates for positioning, communication, and cybersecurity to enable more effective evaluation, together with a formalized operational design domain (ODD) for improved coverage and traceability; (2) multi-concern development management; demonstrating that co-engineered safety and security processes enhance completeness, supported by structured argument patterns, unified process models, and synchronization mechanisms to align assurance activities; and (3) independent assessment methods; developing a complementary testing approach called Assessment of Cybersecurity-informed Safety (AoCiS) and a systematic method for allocating test cases (METAFODD) to appropriate environments, enabling scalable scenario-based validation. Case studies and demonstrations show gains in efficiency, clarity, and completeness, while large-scale validation remains a challenge to be addressed in the future. Overall, the research presented in this thesis establishes structured and transparent methods to support independent and systematic assurance of safety and security in future evaluation efforts.