An efficient method of combining detection and identification of seafloor objects using Gavia AUV

This paper describes a method of dynamically searching for objects on the seabed using side-scan sonar images. Traditionally this has been a two step process where first the area is mapped and then later objects are chosen for inspection in a second pass. In the paper a model is described to explore different search tactics. Using this model we show that it can be beneficial to use an on-line detection algorithm and inspect objects as soon as they are detected. This is because the vehicle will position the objects more accurately and we will not need to travel again through the whole area. The benefit will depend on the number of objects that are in the search area and the quality of the online detection algorithm. An CAD/CAC algorithm was developed as a part of the project and simplified inspection patterns are also explored.