A Taxonomy and Survey on Validation Approaches for Automated Driving Systems

Abstract

The absence of the human fallback level in Automated Driving Systems (ADS) results in a tremendous increase in system complexity and required reliability. Due to the number of possible parameter variations in a driving situation, the framing of all interactions into requirements is infeasible. Therefore, the determination the level of maturity using requirements-based methods is not sufficient anymore. To face this challenge, scenario-based testing is a promising approach. For the actual task of testing, there are a multitude of different test environments, platforms and validation approaches that can be used. Thus, the challenge arises to orchestrate the test efforts among these different possibilities. In this contribution, we present a broad survey on established validation approaches as well as research in this area. Based on the survey, a comprehensive taxonomy for classifying validation approaches is given. The presented taxonomy provides a structured view on available validation approaches and can serve as a tool that supports the selection of validation approaches considering given constraints.