Conceptualising automated driving shared control hazard causes

The motivation for this research was the realisation that the introduction of greater vehicle automation would not only change the task of driving but would also potentially change how vehicles are developed and safety is assured. Undertaking a practice-based workshop identified many Automated Driving (AD) safety assurance challenges having different levels of human-machine control. These challenges include an increase in the size and complexity of AD safety analyses, a need to re-examine the notion of controllability in the context of shared control, and the need to conceptualise the vehicle as a system of systems.

To begin addressing these challenges and to answer the research question “how can the safety of AD be assured under different levels of shared control?” this research has created three products: a vehicle model and behavioural competency taxonomy that allows AD shared control to be conceptualised, a concrete hazard analysis method for analysing AD shared control hazard causes, and a safety case argument pattern for that.

A series of case studies evaluate the research products described above. These cases have used contemporary AD vehicle features, having varying levels of automation. The evaluation of driver assistance, partial and conditional automation cases have been completed by the author. Complementing these is the analysis of a highly automated vehicle system, which has been undertaken with the engineering team from Oxbotica. Considered together these case studies establish the research products as a proof-of-concept hazard analysis method for AD shared control. Further evaluation work is needed to test the viability of the method as an engineering tool for use by automotive practitioners working in a product development environment.