Requirements Modeling and Automatic Transformations Towards Autonomous Driving Scenarios Description

Autonomous Driving Systems (ADSs) have been developed quickly in recent years, robust and complete system requirements are essential for high-quality ADSs. Requirements modeling is a process of describing and analyzing system requirements using various techniques, which is a critical step for ensuring high-reliability requirements. Restricted Use Case Modeling (RUCM) is a use case modeling method to reduce the ambiguity of requirements written in natural language, which has been extended to diverse applications. By extending RUCM with autonomous driving concepts, RUCM for Autonomous Driving Systems (RUCM4ADS) can specify autonomous driving scenarios in the form of use case specifications and diagrams. However, due to the diversity and complexity of autonomous driving scenarios, the constructed requirement models require multiple manifestations to be compared and analyzed. To this end, in this paper, we extend the metamodel of the Unified Modeling Language (UML) class and activity diagrams according to the Wise Drive framework for accurately expressing autonomous driving concepts and behaviors. Further, we realize modeling autonomous driving scenarios with extended UML meta-models, which enrich the manifestation of autonomous driving requirements. Meanwhile, we propose automatic model transformation approaches to promote the automatic analysis of autonomous driving scenario requirements, which generates UML extended models from RUCM4ADS models based on proposed extended metamodels. Finally, two real-world autonomous driving scenarios are used to verify the feasibility and effectiveness of the proposed model transformation approach. Results show that our proposed approach is effective, and can generate high semantic and syntactic accuracy class diagrams and activity diagrams with autonomous driving concepts and behaviors.