Challenges and research approaches for the supra-regional validation of a scenario catalogue for automated driving functions

This admittedly somewhat unwieldy title basically describes the core content of my doctorate, which I was able to present and discuss at the WKM Symposium in Karlsruhe in mid-July. In the following I will present the content of my presentation on the basis of my submitted abstract, but I am always happy to hear more detailed questions and discussions – also by email to my professional address.

Accepted abstract:

Driver assistance systems are becoming increasingly important in the automotive sector and represent a key technology on the road to automated driving. Since the systems have to operate in complex traffic situations, comprehensive testing is necessary for their development and homologation. A means of testing is provided by test scenario catalogs derived from real traffic events, which attempt to depict road traffic events as comprehensively as possible with the aid of normal driving, critical and accident scenarios.

Natural driving data (NDS studies) as well as police and in-depth-accident databases can serve as databases for the test scenario catalogues, like e.g. in the “SePIA” project. These databases represent a sample from a clearly defined population (often highly localised) in terms of time, subject matter and space. Test scenario catalogues generated from the databases are therefore initially very limited in their scope of application.

This is due to the diversity of traffic events with regard to environmental (e.g. topographical), economic (e.g. income-specific) and social (e.g. socialisation of road users) aspects as well as with regard to traffic itself (e.g. volume and composition of traffic). In order to apply the test scenario catalogue for the validation of driver assistance systems as widely as possible (nationally / internationally) as possible, it is therefore necessary to determine its relevance and significance for other regions. This ensures that the diversity of traffic events is taken into account. The methodology used for this should be theoretically simple and quickly applicable to as many regions as possible and offer a high degree of reliability. This is the only way to take account of the rapidly changing global traffic situation and ensure the success of a test scenario catalogue.

However, method development poses several challenges: On the one hand, the requirements mentioned result in a conflict of objectives (to develop a method that is as simple as possible but at the same time accurate and universal); on the other hand, the databases that are actually used often do not correspond to the same basic population and usually arise under different objectives and statistical survey plans. Furthermore, supra-regional comparative databases are not available across the board and their comparability and quality fluctuate. These challenges as well as research approaches for addressing them are part of the presentation.

Information about the novelty value of the work:

The development of a method for the reliable validation of a scenario catalogue generated from inhomogeneous databases (bottom-up) represents a novel research project.

Leave a Reply

Your email address will not be published. Required fields are marked *