Comparing two systematic approaches for testing automated driving functions

Abstract

Thoroughly validating and verifying automated or autonomous driving functions is inevitable for assuring to meet quality criteria for safety-critical systems. In this paper, we discuss two system testing techniques that have been already used for detecting critical situations for the automated emergency braking function based on vehicle simulations. In particular, we introduce combinatorial testing and search-based testing techniques and compare them. Whereas the first is for identifying interactions of parameters that lead to harmful situations considering predefined value domains, the latter is for finding parameter values that cause such critical situations. We discuss the underlying foundations behind the methods as well as their potential application areas. In addition, we summarize the results obtained when using these methods for testing automated emergency braking.