Stochastic Prediction of Lane Change Trajectories for Adaptive Cruise Control

This paper presents a stochastic model for motion prediction of vehicles on the motorway. The predicted trajectories can be used for predictive control algorithms of Advanced Driver Assistance Systems such as Adaptive Cruise Control. The model uses as input actual measurements from the vehicles's radar and camera sensor. In order to deal with the prediction uncertainty, a graphical modeling approach is proposed that allows to incorporate the turning indicator signal of a traffic participant. The model is trained and evaluated with real measurements. The potential benefits of such a prediction model are demonstrated for the application of Adaptive Cruise Control where the incorporation of the predicted trajectories lead to a significant improvement of safety and fuel efficiency.