A Rigorous Analysis Template Process to Capture the Safety Properties of Self-Driving Vehicle Systems

Abstract

Self-Driving Vehicles (SDVs) are seen as a significant advancement in the automotive domain, hinting at a future where human drivers might be rendered obsolete. However, even with the advancements in SDV technology, the need for human drivers is still recognised. The incorporation of human drivers into SDVs introduces unique and significant challenges. The significance of human driver and SDV interactions cannot be overstated, especially when the SDV relies on the human driver as a fallback option during hazardous driving events. To address this critical aspect, this thesis presents a methodology termed the Rigorous Analysis Template Process (RATP).

RATP establishes an analytical journey to develop a comprehensive framework ensuring safety and optimal cooperation between human drivers and SDV systems. It represents an evolution in existing work on analysing system safety and provides a more rigorous systematic strategy for SDV systems. It involves both systematic analysis and formal methods to evaluate safety in SDV systems.

Drawing strength from a combination of both systematic analysis and formal methods, RATP adeptly identifies high-level safety requirements and develops a rigorous model to investigate issues and assumptions that may arise during the operations of SDV systems. One of the key benefits of RATP is its modularity, offering researchers and developers the ability to systematically analyse system behaviours from a high-abstraction view down to a more detailed view. The conclusion of this research presents a robust set of modelling patterns that act as a blueprint for the future development of SDV systems.

RATP is demonstrated with a case study that explores the various functionalities of an SDV system to evolve the methodology into a mature state. Finally, this thesis presents a discussion on future improvements that could be undertaken to develop the methodology further.