Software Architectural Design for Safety in Automated Driving Systems
Self-driving cars have generated a revolution in the automotive industry. Nonetheless, ensuring safety in the absence of a supervising driver and verifying safe vehicle behaviour in various contexts are two of the main challenges for autonomous driving systems that need to be addressed in the near future. Due to their complexity, Autonomous Driving Systems (ADS) cannot be solved in a straightforward way without being properly structured. Therefore, it requires a well-defined architecture to guide its development. In addition to providing modularity and scalability, the proper architecture provides a maintainability system. To help overcome some of the challenges, this master thesis develops an architecture for an ADS that adapts its behaviour to the context by switching between different operational modes, with the aim to standardize and ease the development process. The work was divided in four parts. First, the safety standards for the development of an autonomous functions have been analysed. Second, the system’s requirements were derived from a widely adopted automotive standard. Third, a logical architecture has been proposed and instantiated for an automated parking system. Finally, the architecture has been implemented in a simulation environment for its proper validation. This work has shown that the architecture modelled in AUTOSAR and the generated Run-Time Environment is capable of adapting its behaviour to the context by executing the mode switch. In addition to meeting the safety requirements for a safe autonomous parking system. The interface created with the simulation environment allows future works to benefit from it for the development and testing of actual developed autonomous systems.
Barcelona, Univ., Master Thesis, 2022
Bayerisches Staatsministerium für Wirtschaft, Landesentwicklung und Energie