A Generic Concept of a System for Predicting Driving Behaviors

Today, many vehicles are equipped with Advanced Driver Assistance Systems (ADAS) to warn the driver about the potential danger of a scene, but in some situations the warning is not early enough to avoid an accident. A solution for preparing the driver and giving him the time to react to such dangerous events is to predict the behavior of other traffic participants. This paper describes a method to predict the behavior of the surrounding vehicles by a classification approach. However, the behavior alternatives strongly depend on the scenario faced by the target vehicle. Where most of the state-of-the-art approaches focus on a single scenario, the concept presented in this paper aims at a generic solution, allowing for behavior prediction for a large amount of different scenes. The idea of the method is to categorize scenes into a hierarchy from the most generic ones in the top nodes to the most specific ones in the leaves. Every node of the hierarchy is a scene containing a set of classifiers to predict the possible behaviors. GPS and digital maps provide the static information about the infrastructure, which is used to determine the nodes fitting to the current situation. As a first step this paper shows accurate prediction of traffic participants behavior in highway entrance situations for a prediction horizon of up to 3 seconds.