Bringing automated driving software to market is a challenging task. Mainly because complex software needs to be developed and verified in a potentially unlimited driving scenario space. Although alternatives exist, academia and industry are focusing on the scenario-based approach, which aims at evaluating sample scenarios that belong to a specified Operational Design Domain (ODD). Significant effort is spent on finding optimization techniques to detect corner-cases and reduce the number of scenarios required for verification. Loss functions for optimization are mostly safety-related and not directly linked to large parts of the specification of the driving function. Consequently, large parts of the specification play no role in the corner-case detection, reducing the effectiveness of the whole verification process. To address this drawback, this contribution applies principles from Behavior Driven Development (BDD) to create this link. We conducted a case-study in which we described the requirements of an automated driving function on system level using Gherkin syntax and used simulation to evaluate the requirements of the driving function with regard to the specification. Our results show that Gherkin syntax helps to overcome communication gaps between system engineers and test engineers. Furthermore, logical driving scenarios do not have to be tailored to the test goal, hence, scenarios remain function-independent. We found that scenarios can be reused in different development branches and Gherkin files can be reused in different verification approaches. Finally, we conclude that using Gherkin syntax to describe automated driving functions is a promising approach to link large parts of the system specification to state-of-the-art formal verification and optimization techniques.
Benefits of Behavior Driven Development in Scenario-based Verification of Automated Driving
08.10.2022
391967 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
MODEL BASED SCENARIO SPECIFICATION FOR DEVELOPMENT AND TEST OF AUTOMATED DRIVING FUNCTIONS
British Library Conference Proceedings | 2016
|