Abstract
LiDAR plays an important role as external sensor for automated driving system. However, low-resolution 3D-LiDAR is insufficient that the point cloud data become sparse at long-range. This problem makes it difficult to obtain accurate information about the target object. To solve this problem, we propose an interpolation method, using sensor fusion with camera and LiDAR. And, RGB data from camera is used to search for corresponding point in adjacent frames. In addition, the searching range is determined by LiDAR’s depth data. We evaluated computing cost and shape reconstruction accuracy while applying our method into preceding vehicle. Furthermore, the point cloud data are carried out clustering and classification based on SVM.
