State of Research on Data-Driven Safety Assurance Methods

Authors

Nicolas Wagener, MSc, Hendrik Weber, MSc, Julian Bock, MSc, Univ.-Prof. Dr.-Ing. Lutz Eckstein, Institute for Automotive Engineering (ika), RWTH Aachen University;
Dr.-Ing. Adrian Zlocki, fka GmbH, Aachen

Summary

To introduce highly automated driving functions into the market, the validation and verification of these functions is a key prerequisite. Established approaches for safety assurance of ADAS and driving functions are, however, not feasible for higher levels of automation. Thus, new technologies and methodologies are necessary to effectively and efficiently assure the safety of highly automated driving. Within this paper, a datadriven scenario-based methodology is discussed, which aims at providing a solution to the open question on safety assurance. The key principle of such methodologies is a database of field data scenarios from various data sources. To give an overview on current activities, this paper presents state-of-the-art projects on data-driven safety assurance methods. Moreover, the common methodology, merged from state-of-the-art approaches, is presented in detail. Different types of data sources will be introduced and discussed. Furthermore, a methodology for the definition of scenarios is described, concluding with a short introduction on the execution of tests. After that, a more in depth analysis is introduced by presenting an exemplary implementation of a database to generate relevant scenarios for the validation of highly automated driving functions.

Members Login