Macroscopic Safety Requirements for Highly Automated Driving in Urban Environments

The release of automated vehicles is only possible if their safety is ensured. Thus, acceptable safety requirements for all stakeholders have to be defined. The scope of this work is a traceable proposal of a definition of macroscopic safety requirements for highly automated driving in urban environments. Firstly, relevant stakeholders are identified. These are the society in general, the users of automated vehicles and other road users, which are further divided into subgroups, depending on their type of traffic involvement. Secondly, the available statistical data for Germany is analyzed. Hereby the domestic annual mileage is identified as the most suitable parameter to relate the stakeholder-specific accident numbers to. The quotients of accidents of different stakeholders and the driven distance represent the current risks of manual driving in urban environments, which are used as exemplary safety reference. Thereafter, existing risk acceptance principles from other domains such as the railway industry are used to determine the acceptable risks for each stakeholder group, which can be directly used as macroscopic safety requirements. Among other analytics, it is shown that the users of automated vehicles are attributed to the highest risk tolerance, while other road users are the most critical stakeholder group in terms of safety requirements.