This dissertation develops a methodological approach for selecting representative concrete scenarios from logical scenarios in order to efficiently validate and approve highly automated driving functions. It presents a multi-stage selection process that combines combinatorial exploration, scenario reduction, and similarity and uncertainty analyses to reduce the testing effort while enabling a robust assessment of the coverage of the scenario space.

The target is to achieve a valid generalization to untested scenarios using a limited number of test cases, thereby demonstrating the safety and validity of automated driving functions in an open-world context.