Using Generalised Self-Organizing Maps as Part of Underwater Localisation for Quay Wall Inspections

• Published in: Global Oceans 2020: Singapore – U.S. Gulf Coast pp. 1 - 8
• Main Authors: Ehlers, Kristian, Isokeit, Cedric, Meyer, Benjamin, Behrje, Ulrich, Maehle, Erik
• Format: Conference Proceeding
• Language: English
• Published: IEEE 05.10.2020
• Subjects: AUVs; Inspection; Inspections; Quay Walls; Security; Self-organizing feature maps; Self-Organizing Maps; Sonar detection; Task analysis; Training data; Underwater Localisation; Underwater structures
• DOI: 10.1109/IEEECONF38699.2020.9389242

The length of the quay walls of the port of Hamburg, which is the largest seaport in Germany and the third largest in Europe, is about 43km. The facilities and installations, including the 147 bridges, are subject to constant corrosion and must therefore be periodically inspected for their condition to ensure safety in the port. The majority of the work is done by divers who use their hands and years of experience to inspect the underwater structures for damage. As this type of inspection can only be carried out on a random basis, Autonomous Underwater Vehicles (AUVs) will offer a promising tool in the future to cope with the demands in a port environment. A stable localisation is the basis for inspection tasks and the subject of many publications. This paper describes a novel approach of using a generalised Self-Organizing Map for underwater wall detection based on sonar data enabling a faster and more robust localisation technique without prior recording of training data.