Software Architectural Design for Safety in Automated Parking System

The automotive industry has seen a revolution brought about by self-driving cars. However, one of the main challenges facing autonomous driving systems is ensuring safety in the absence of a supervising driver and verifying safe vehicle behaviour under various circumstances. Autonomous Driving Systems (ADS), due to their complexity, cannot be solved straightforwardly without proper structure. Thus, they need a well-defined architecture to guide their development with requirements that involve modularity, scalability, and maintainability among other properties of a well-defined architecture. To help overcome some of the challenges, this master thesis defines and implements in a simulated environment an automated parking system that complies with industrial and safety standards. The work has been divided into four parts. Firstly, the safety rules for the development of an autonomous function have been analysed. Secondly, the use cases and system requirements have been defined following the needs of the automated parking system. Thirdly, the system has been implemented in the simulation environment with a structure based on a widely adopted automotive standard. Finally, the final result is the software architecture of an autonomous vehicle with automated parking functionality. This concept has been validated within the virtual environment together with the integration of the AUTOSAR run-time environment, which is in charge of executing mode functionality switching and CARLA which is the simulation environment. This project has proved that based on safety standards it is possible to design and implement new functionalities in safety-compliant autonomous parking systems both using stablished technologies and new ones. This final result allows future work to benefit from the integration between architecture and simulation, thus easing the development and testing of future autonomous systems.