Clustering of Detected Targets Using DBSCAN in Automotive Radar Systems

In this paper, we propose a method of clustering detected targets using density-based spatial clustering of applications with noise (DBSCAN) in the automotive frequency modulated continuous wave radar system. When the DBSCAN method is applied to target detection results, detected targets having similar velocities, distances, and angles are grouped into the same cluster. Through this clustering process, we can focus on only the desired target in the field of view of the radar system by removing trivial clusters corresponding to road structures such as guardrails and iron pillars. In addition, we can approximately estimate the road curvature by utilizing the detection results of road structures. The proposed target clustering method can provide useful information to radar-equipped vehicles using adaptive cruise control and autonomous emergency braking functions.