Application of Model-Based Systems Engineering Methods in Virtual Homologation Procedures for Automated Driving Functions

Further development of vehicles into software-defined products is increasingly influencing vehicle approval [1]. In particular, approval procedures are required for updates to homologation-relevant software, such as automated or autonomous driving functions. Appropriate solutions such as the "Digital Loop" showcase [2] focus primarily on virtual homologation methods. The verification in such procedures requires complete traceability from legal requirements to adjustments in the software architecture and functionality as well as to derived test cases and the tested digital twin. When applied to automated driving functions, scenario-based methods and a high degree of automation due to the scope of testing are also required. This contribution uses the automated driving function Automated Lane Keeping System (ALKS) to show how software extensions can be formally specified using the scenario and model-based systems engineering method CUBE based on legal requirements. The scenario, requirements and architecture specification using Systems Modelling Language (SysML) can then be used for the implementation in the vehicle and in the Digital Twin as well as for homologation. In addition, the automatic derivation of machine-readable, homologation-relevant test scenarios and test cases from the specification model is shown. The test specification generated in this way can be used directly for simulation-based tests and vehicle tests. The use of the procedure in the homologation of software updates enables complete traceability with reduced specification and testing effort at the same time.