Improving the Modelling of the Lateral Driving Behavior of Vehicles within their Lane Over Extended Time Periods

The lateral movement of vehicles within their lane significantly influences the range of vision of sensors automated driving functions rely on. Consequently, it is crucial for the objects detected by downstream object detection algorithms. These in turn are decisive for the situational awareness of an automated driving function. Due to the rising importance of simulations for the validation of automated driving functions, it is necessary to reflect the lateral movement of vehicles within their lane in simulations. For this, so called submicroscopic behavior models have been developed. These, however, show limitations when it comes to the modelling of lane following maneuvers in the range of half a minute up to a few minutes. The present paper addresses this issue: based on an analysis of the driving behavior on this scale in real-world data, several approaches are proposed to better reflect the identified effects within the model introduced by the authors in earlier work. The results of these adaptions are qualitatively and quantitatively evaluated. Then, the different approaches are categorized according to their potential use cases, and future research directions are derived.