Eine agile Methode zur simulativen Qualitätssicherung von aktiven Sicherheitssystemen

Kühnel, Stefan; Rumpe, Bernhard (Thesis advisor); Berger, Christian (Thesis advisor)

Düren : Shaker Verlag (2022)
Book, Dissertation / PhD Thesis

In: Aachener Informatik Berichte Software Engineering 51
Page(s)/Article-Nr.: xvii, 317 Seiten : Illustrationen, Diagramme

Dissertation, RWTH Aachen University, 2021

Abstract

Context: In addition to the current efforts to advance the electrification of the powertrains through innovations, the integration of driver assistance systems in the context of automotive development is of particular importance, both to increase driving comfort and to improve vehicle safety. Consumer tests, e.g. from organizations such as EuroNCAP, play an important role, which serves both as a key driver for safety improvement and to assess the performance of the underlying safety systems. In order to be able to better assess and increase the quality of the software developed for this purpose, tests both in designated test areas under real conditions and simulation-based tests in synthetic environments are suitable approaches to meet the challenges of continuous quality improvement in an agile way, although both approaches have different hurdles and limits. In particular, the simulative approach considered here often leads to the dilemma that the development of suitable simulation environments also requires a high amount of resources like the actual system development. As a result, this continues to progress in parallel and the simulation environment cannot be used in adequate time due to the high effort. Such apossibility can occur, for example, if the aspect of modeling with the quality goal of being as realistic as possible is defined without feedback to the question at the beginning: "What should be answered by the simulation?". Goal: The main goal of this thesis is to support the complex development process of active safety systems in the context of consumer tests with a simulative approach to improve software quality. Furthermore, the allocation of resources during development and the necessary real tests should be improved and thus made more effective through the behavioral analysis of algorithms. Method: After carrying out a Systematic Literature Review (SLR) to examine any existing approaches and methods for the development of such simulation environments, a separate method is developed, presented and applied in a case study, taking into account the existing project conditions. Results: The analysis of the project context comes to the conclusion that there are simulation activities, but a structured approach to their development is missing. The Systematic Literature Review confirms this result, so that the need to develop an agile method for simulative quality assurance of active safety systems, especially with regard to consumer tests, is shown. The method presented comprises four building blocks: (a) the analysis and modeling of the investigation area, (b) the development of an adequate simulation infrastructure, (c) the development of evaluation methods and (d) the implementation of simulation runs and their evaluation. Finally, the method is applied with the help of a case study for a proof of concept. Conclusion: It is shown that the method makes a positive contribution during the qualitative assessment of software components on a simulative basis, especially where equivalence class tests are not sufficient to adequately test consumer test scenarios. Limitations and the need for expansion of the method are primarily seen in the transfer to other contexts in the driver assistance environment and the expansion to additional consumer test scenarios such as pedestrian protection.