An algorithm to identify docking locations for autonomous surface vessels from 3-D LiDAR scans

摘要

In this paper we present a novel algorithm to identify docks (or piers) from Light Detection and Ranging (LiDAR) scans. The algorithm exploits the expected geometric features of the dock and does not require modifying the dock in any way. Our approach consists of a novel combination and application of open source tools for point cloud and image processing. In our limited testing on 8 fused data sets, the algorithm successfully identified a usable portion of all the docks with only one false positive (a seawall). The target application is automated docking (a.k.a. recovery) for small Unmanned Surface Vessels (USVs).