UAV-aided Large-scale Map Building and Road Extraction for UGV

Due to the limited visual field of UGV, it is difficult to detect the accessible zone in a wide range of area in real time. This paper proposes a UAV-aided autonomous road network construction method where roads are previously unknown for UGV navigation. First, SIFT and RANSAC algorithms are applied to obtain the homography of UAV sequential images, through which these images are stitched to a large-scale map. Next, least square fitting algorithm and RDP algorithm are applied to extract initial road area and then obtain road network suitable for UGV navigation by topological processing and skeleton extraction. The experiments show that this UAV-aided method is effective in generating a large-scale road network map for UGV driving through unknown environment without human intervention.