Objective Controllability Assessment for Unintended ADAS Reactions

Abstract

For the development of Advanced Driver Assistance Systems, the assessment of benefits and potential risks are crucial factors. An example is the controllability assessment of unintended reactions of ADAS as described in ISO 26262. One method this standard proposes for the assessment is the car clinic with naïve subjects. The effort needed to apply this method is analyzed regarding the statistical boundaries. As this effort is considered to be unjustifiable in most cases and in particular if the transferability of results to other systems or situations is low, an approach to increase the transferability is outlined. Therefore, controllability situations due to unintended reactions of ADAS functions are analyzed for contributing and influencing factors, differentiating between the driver and the environment. The goal is to identify a minimum set of necessary test cases with high situational relevance and high influence on controllability. To measure the change in controllability, an assessment criterion for a hazardous situation in longitudinal traffic is developed. The combination of the situational relevance and the controllability assessment allows an overall relevance factor to be derived which weights the situations. This enables the identification of the necessary test cases for assessment of unintended reactions of Advanced Driver Assistance Systems.