Ensuring proven safety is the biggest limiting factor for automated driving. New distributed vehicular architectures, shifting from traditional ECUs to powerful centralized computers, bring flexibility and cost efficiency — but also unprecedented complexity. Highly redundant and complementary sensor setups must deliver reliable perception across diverse conditions, while classic validation methods fall short. This presentation introduces RT-CARS, an innovative tool for holistic modeling and Monte Carlo-based reliability analysis, taking into account software and hardware dependencies, diverse failure and repair modes, fault tolerance and resource management. Its capabilities will be shown on behalf of a published, real-world architecture of an autonomous vehicle.

• Key aspects and challenges on the validation and certification processes of autonomous vehicles
• Identification of architectural aspects relevant for the safety modeling and analysis
• Limitations of traditional validation methods
• Modeling and analysis capabilities of RT-CARS and simulation insights
• Utilization of the evaluation results