Data fusion and assessment for maneuver prediction including driving situation and driver behavior

With the rising number and complexity of advanced driver assistance systems the need of adapting the assistance depending on driver intentions and driving situations becomes indispensable. But for this, vehicular systems have to be enabled to detect and assess driving situations and to derive driver intentions in real-time and as soon as possible. Therefore two sources of information can be used - environmental recognition and driver observation. In this paper a set of features assessing the environmental situation, assessing the driver behavior and a method integrating both aspects for predicting lane change maneuvers are proposed. The method is implemented as a real-time application and is applied on naturalistic driving data.