Radar-based extended object tracking under clutter using generalized probabilistic data association

An important foundation for various vehicular applications is a reliable environment recognition. In this context, the simultaneous estimation of the state and the existence of an unknown number of objects under difficult detection conditions is a particular challenge. In this paper, we propose an algorithm for tracking extended objects under clutter. We propose an extended measurement model which enables the estimation of the object width using a standard Kalman filter implementation without the need for clustering the data. As this implies multiple observations generated by one object and additional clutter observations, the generalized probabilistic data association with a state-depended cardinality model is utilized. The proposed algorithm is evaluated with simulated data of a radar-based vehicle tracking system.