Risk Assessment for Integral Safety in Automated Driving

This paper presents a novel risk assessment approach that allows to control the behaviour of self driving cars. This novel continuous real-time risk assessment considers uncertainties as well as accident severity predictions to intervene integrally. Thus it not only allows predictive traffic interaction and collision avoidance, but also an intelligent crash interaction. These decisions are made on incomplete data, due to imperfect environment perception data and road users' unknown intentions. Advanced, situational and numerical dependencies are regarded. Furthermore, the benefit of multiple approximating accident severity estimations are discussed.