Towards the Automation of Visual Inspections of Cargo Holds of Large-Tonnage Vessels

Seagoing vessels have to undergo regular visual inspections in order to detect the typical defective situations affecting metallic structures, such as cracks and corrosion. These inspections are currently performed  by ship surveyors manually at a great cost. This paper describes a \emph{Micro-Aerial Vehicle} (MAV) intended for the visual inspection of cargo holds, whose development, among others, takes place within the context of the EU-funded H2020 project ROBINS, aiming at making ship inspections safer and more cost-efficient. The vehicle is equipped with specific sensors that are to permit teleporting the surveyor to the areas that need inspection. The focus of the platform control software is on providing enhanced functionality and autonomy for the inspection processes. All this has been accomplished in the context of the supervised autonomy paradigm, by means of the definition of different autonomy levels and functionalities (including obstacle detection and collision prevention), and extensive use of behaviour-based high-level control, all intended for visual inspection. Automatic detection of defects is also addressed as part of ROBINS goals, through the adoption of deep learning approaches for enhanced performance. Results for some experiments conducted to assess the different functionalities are reported at the corresponding sections of the paper.