The dissertation develops a methodological approach for the selection of representative concrete scenarios from logical scenarios in order to efficiently design the validation and release of highly automated driving functions. A multi-stage selection process is presented that combines combinatorial exploration, scenario reduction as well as similarity and uncertainty analyses in order to reduce the test effort and at the same time enable a reliable statement about the coverage of the scenario space.

The aim is to achieve a valid generalization to untested scenarios with a limited number of test cases and thus demonstrate the safety and validity of automated driving functions in an open-world context.