Bayesian Visual Tracking for Inspection of Undersea Power and Telecommunication Cables

Abstract

The surveillance and inspection of underwater installations such as power and telecommunication cables are currently carried out by trained operators who, from the surface, guide a Remotely Operated Vehicle (ROV ) with cameras mounted over it. This manual visual control is, however, a very tedious job that tends to fail if the operator looses concentration. This paper describes a tracking system for underwater cables whose main objective is to allow an Autonomous Underwater Vehicle (AUV) to video-document the whole length of a cable. The approach is based on Particle Filters (PF) because of their natural ability to model multidimensional multi-modal probability density functions, what allows handling in a more appropriate way the ambiguities which naturally arise from undersea environments. Extensive experimental results over a test set of more than 10,000 off-line frames, for which a ground truth has been manually generated, have shown the usefulness of the solution proposed. All those images come from inspection runs captured by ROVs navigating over real power and telecommunication undersea cables. Besides, on-line results obtained from an unmanned vehicle guided by the cable tracker in a water tank are also available and are discussed in the paper.