Multiview Fusion Automotive Radar SLAM

A multiview fusion automotive radar simultaneous localization and mapping (SLAM) method is proposed to solve the problem of automotive radar point cloud mismatch caused by targets’ radar cross section (RCS) glint. The proposed method suppresses the RCS glint by fusing the images from multiple radars with different views. Meanwhile, to fully exploit the intensity information provided by the radar point cloud, a radar point cloud matching algorithm is proposed, which significantly improves the point cloud matching accuracy. The effect of factors, such as the number of radars and the radar layout on the suppression of RCS glint, is also analyzed. Finally, radar SLAM experiments were conducted in underground garages and outdoor roads. It is verified that the proposed method can improve the localization accuracy by an order of magnitude compared to existing radar SLAM methods, providing decimeter-level localization accuracy in seconds and constructing maps that are consistent with the real environment.