The worst-time-to-collision metric for situation identification

Currently, the introduction of highly automated vehicles is one of the major targets of the whole automotive industry. However, it is still unclear how to cope with the testing effort necessary to approve an automated vehicle. One possibility to reduce the testing effort is to focus the assessment on critical situations. To describe the criticality of these situations, metrics are required. Firstly, this paper states requirements on assessment metrics. Secondly, this paper introduces a simple but comprehensive metric to select objects and situations out of a typical test drive to reduce the amount of data saved for further analysis. As it must not be assumed that the same situations are critical for human drivers and for automation, the metric only relies on driving dynamics and the physical possibilities of the vehicle. The special feature of this metric is the worst case assumption for vehicle behavior. If a situation is uncritical, even with the worst possible maneuvers, it is allowed to be neglected in the assessment process.