The safety assessment of driving functions poses a key challenge on the way to highly automated driving. In the PEGASUS project, fka already participated in the development of a method for storing relevant features of traffic scenarios in a generic database. A follow-up research project covers the evidence-based derivation of basic scenarios for controlled-access highways. Data sources employed by fka in this context include accident data, field operational tests, driving simulator studies and in particular traffic observation using infrastructure sensors and drones. As all sources constitute just singular measurements, a complete acquisition of all traffic is not viable, though.

The goal of the BASt project (FE 82.0735/2019) “Concept for Statistically Representative Traffic Observations” is to develop a concept for the representative gathering of scenarios in traffic areas like federal highways as well as urban and rural areas. A sample taken on the basis of this concept can be used to derive statistically valid conclusions about the frequency of individual scenarios in the context of the traffic in total or a specific traffic domain. The result of this project constitutes a first approach for a proof of completeness and therefore is a central building block for an efficient and effective approval process of automated driving systems.

Approach

First, different approaches for scenario concepts aiming to structure and describe scenarios within the traffic domain are examined and a universal description is derived. The relevant traffic scenarios as well as the context factors influencing their characteristics and frequencies are identified. Statistical measurements of these factors’ frequencies facilitate the establishment of a methodology to generate sampling plans for the collection of representative scenarios and their subsequent extrapolation to the traffic in total. This methodology is to be analyzed and discussed in an upcoming expert workshop. Finally, an exemplary sampling plan is generated and realized to validate the concept.