Vision based localization system for AUV docking on subsea intervention panels

The last generation of autonomous underwater vehicles (AUV) is being developed with intervention capabilities in mind. These capabilities will provide easy and early inspection and maintenance of subsea offshore structures. The technology developments that implies this upgrade comprise new techniques for subsea vehicle localization, including detection and estimation algorithms. A reliable technique for robust localization of an underwater vehicle on the boundaries of a subsea intervention panel, based on computer vision as well as traditional navigation sensors, is presented. The prior knowledge of the structure where the vehicle has to dock on, allows to use model based pose estimation techniques that rely on a list of detected features. The system is mainly based on a sensor fusion scheme that produces pose estimation in 6 degrees of freedom using the information from different sensors and the vision module.