Toward Real-Time Visually Augmented Navigation for Autonomous Search and Inspection of Ship Hulls and Port Facilities

This paper reports on current research to automate the task of ship hull inspection and search using autonomous underwater vehicles (AUVs). We describe an automated feature-based navigation (FBN) and mapping framework that provides the AUV with precise in-situ hull-relative localization. Our vision-based perception approach eliminates the need for having to deploy additional navigation infrastructure, such as acoustic beacons (a traditional method for obtaining precise bounded-error navigation). We describe our mapping framework and show how we are now applying that framework to the task of automated ship-hull inspection using the HAUV testbed. The operational impact to the Navy of this technology development will be rapid, repeatable, automated 100% survey coverage for ship-hull inspection.