Identifying relevant traffic situations based on human decision making

Autoren

Christoph Sippl, Florian Bock, Bernd Huber, AUDI AG Ingolstadt;
Anatoli Djanatliev, Reinhard German, Friedrich-Alexander Universität Erlangen-Nürnberg

Jahr

2019

Zusammenfassung

The growing complexity of future driving systems and the shorter development cycles pose huge challenges in developing and validating automated driving functions. User expectations and respective requirements must be fully described in detail to enable a complete specification of the target system and deriving test cases for the validation process. The goal of automated driving functions is to assume the driving task not only in specific situations as driver assistance systems, but to handle all upcoming situations in a defined traffic environment. The consequence is a rising complexity of the target system to recognize the environment, interpret the perceived situations and its elements, as well as making adequate decisions for the upcoming driving task. The identification of all relevant situations which have to be managed by the driving function is not possible with the available methods. Thus, requirements for automated driving functions can only be described in an abstract way, which leads to difficulties in the safety validation process. Bergenhem et al. state this as semantic gap.
Scenario-based methods for the development and the validation of automated driving functions are frequently discussed in recent research. Scenario-based methods on the one hand are able to describe user expectations or expected system behaviors in an abstract way. Further, they can be used for hazard analysis and risk assessment for driving functions. On the other hand, scenario-based methods are discussed for the validation of automated driving functions, as real test vehicles using route based methods are neither economically justifiable nor being safe. Therefore, scenarios depict the basis for virtual validation using, e.g. X-in-the-loop methods.
The complexity of future driving functions, the challenges in requirement engineering such as the semantic gap and the growing amount of relevant traffic situations make it hard to completely specify the system in a detailed way. Identifying relevant situations and the formulation of scenarios to derive requirements and related test cases is one task to overcome the challenges of requirement engineering, system specification and validation of future driving systems.

Mitglieder-Login