Advantageous Usage of Textual Domain-Specific Languages for Scenario-Driven Development of Automated Driving Functions

摘要

Current driving functions are continuously improved. As target vision of this development, the final state of unattended reliable navigation under all possible circumstances is defined. Besides the requirements and the test cases, the main accompanying specification artifact for the iterative development of such functions is the scenario catalog, which is usually created by hand. The included scenarios often differ in terms of the used abstraction level, the chosen natural language, and the degree of completeness depending on the current development phase. There is currently no methodology available, which allows the iterative evolution of textual scenario descriptions based on natural language and that can be used for the development of automated driving functions (ADF). As a solution approach, the methodology scenario-accompanied, text-based, iterative Evaluation of automated driving Functions (stiEF) as embedding workflow and an embedded textual domain-specific language (DSL) for scenario creation are presented. The DSL facilitates the iterative multilingual evolution of scenario descriptions and the generation of the corresponding simulation parameters and scenario sketches. For this, scenarios with different levels of detail are used as central specification artifacts. This approach avoids ambiguities in the descriptions, guides the user through their creation process, and ensures completeness. This is demonstrated by a prototypical realization.