The Monitor as Key Architecture Element for Safe Self-Driving Cars

While research in self-driving cars continuous to boom, research of their dependability aspects is only emerging. In particular, trade-off studies on fault-tolerant architectures for safe self-driving cars are largely missing, but desperately needed by the automotive industry. In this paper we give a brief overview of some possible architecture designs, and argue in favour of those architectures that implement a monitor element. We also formulate open research questions with respect to monitor construction and permissible interfaces from the monitor to other elements that constitute the architecture.