I. Introduction
When thinking about human driving behavior, it seems to be obvious, that it is not only affected by the current traffic situation, but by various external conditions as well. For example, the weather situation, traffic density or daytime can depict such conditions. Knowledge about external conditions is also used by human drivers for improving their motion predictions of other traffic participants. This context-awareness is one important aspect distinguishing the ability of humans in predicting other vehicles movements from the one of current advanced driver assistance systems. Therefore, our hypothesis is that an improvement of the system’s performance towards a human-like one can be achieved by taking contextual information and especially knowledge about external conditions into account when developing motion prediction systems. Fig. 1 visualizes this thought. In this paper, we are investigating the impacts of external conditions on driving behavior as well as on the performance of current motion prediction systems.