Optimal coverage in automotive configuration

Abstract

It is a problem in automotive configuration to determine the minimum number of test vehicles which are needed for testing a given set of equipment options. This problem is related to the minimum set cover problem, but with the additional restriction that we can not enumerate all vehicle variants since in practice their number is far too large for each model type. In this work we illustrate different use cases of minimum set cover computations in the context of automotive configuration. We give formal problem definitions and we develop different approximate (greedy) and exact algorithms. Based on benchmarks of a German premium car manufacturer we evaluate our different approaches to compare their time and quality and to determine tradeoffs.