Clustering and Subsequent Contour and Motion Estimation of Automotive Objects Using a Network of Cooperative Radar Sensors

A method for collaborative target list processing of multiple radar sensors simultaneously estimating multiple extended targets is presented and discussed. A scheme to cluster target points from multiple radar sensors jointly is described that deals with overclustering and separation of objects. Algorithms to estimate the contour (i.e., dimensions) and motion of these objects are proposed and evaluated. Additionally, the influence of the number of sensors on the results is investigated and compared to single sensor estimates. The processing of measurements with three radar sensors and two vehicles shows the advantages of using more than one sensor, as well as the proposed algorithms.