Consistent mapping of marine structures with an autonomous surface vehicle using motion compensation and submap-based filtering

Abstract

Inspection and maintenance of marine energy farms are becoming increasingly important as more farms are developed and constructed. In this study, structural monitoring system using an autonomous surface vehicle (ASV) has been proposed to enable the inspection of an offshore power plant under harsh ocean disturbances. In particular, the ASV uses guidance and control laws to minimize the deviation from the desired path in waypoint tracking. The ASV was also equipped with multi-modal sensors - sonar and light detection and ranging (LiDAR) - for mapping underwater and surface structures. By utilizing the navigation sensors, the motion of the ASV in ocean disturbances was accurately estimated and incorporated into the mapping process to produce consistent multi-modal maps. Furthermore, dedicated outlier removal methods were proposed to improve the performance over conventional approaches in real offshore data. We considered the sparsity of the sonar and LiDAR data in their respective modality during acquisition and designed novel submap-based filtering to remove the outliers. To validate the proposed system, a field test was conducted on an offshore wave farm structure. It was verified that the ASV can cope with ocean disturbances in its waypoint tracking, and the resulting multi-modal maps are consistent and outlier-free.