Rope augmented path following and control of remotely operated underwater vehicle using vision for stilling basin surveillance

Abstract

The high energy contained in Hydroelectric dam projects leads to the erosion of stilling bucket structures. Failure to maintain the stilling basin can ultimately lead to erosion of the spillway structure and thus affect the entire project. Deep learning methodologies have been advanced for the monitoring of cracks and condition of concrete within large scale engineering projects. Nonetheless, these techniques are generally applied subsequent to data acquisition and necessitate the involvement of human divers to manually collect the data and click photographs. This paper introduces a prospective resolution for the monitoring of the stilling basin by harnessing computer vision algorithms. To accomplish this, an autonomous underwater robot is deployed to conduct a comprehensive assessment of the designated basin and subsequent analysis of the gathered data. Path planning and guidance for the autonomous underwater vehicle are facilitated through rope-augmented navigation techniques, Deep learning techniques are applied on the collected data in real-time to detect the erosion in structure accurately. The project is implemented using robot operating system (noetic) on BlueRobotics BlueRov2 simulated using UUV Simulator in Gazebo Ignition.