Szenario-Optimierung f\"ur die Absicherung von automatisierten und autonomen Fahrsystemen

TL;DR

This paper introduces a metaheuristic search methodology for optimizing test scenarios in automated and autonomous driving systems, aiming to improve verification, validation, and system safety assessment.

Contribution

It presents a novel approach combining scenario parameterization, search space definition, and fitness functions to identify worst-case scenarios for autonomous driving systems.

Findings

Effective identification of critical test scenarios

Enhanced test completeness and system quality assessment

Automated test goal-oriented testing enabled

Abstract

The verification and validation of automated and autonomous driving systems impose a major challenge, especially the identification of suitable test scenarios. This work presents a methodology that adopts metaheuristic search to optimize scenarios. For this, a suitable search space and a suitable fitness function needs to be created. Starting from abstract descriptions of the system's functionality and use cases, parameterized scenarios are derived. The parameters span a search space, in which the suitable scenarios need to be found. Guided by a fitness function, search-based techniques are used to identify those scenarios, in which the system shows its worst behavior. If the derivation of the fitness function is done correctly, an argumentation basis about test completeness and system quality may be achieved. Further, test goal oriented testing with automated test oracles is enabled.