Development of a perception and control module for a remotely operated underwater vehicle performing fish cage inspection
A Blueye PRO Remotely Operated Vehicle (ROV) is being adapted to perform automated inspection of marine fish farm cages. A Robot Operating System (ROS) node should be developed to enable periodic picture taking using the on-board camera of the vehicle, in accordance with the Python SDK provided by Blueye. For the purpose of fish cage inspection, and avoiding rope and vehicle/tether entanglement, certain manoeuvres must be performed underwater. A proposed way of moving around the cages is following the recognisable segments of the outer construction of the fish cage from the surface downwards by controlling the vehicle's yaw, heave, and depth. The boundaries of the net must be detected and the image from the camera must be segmented in order to enable a visual servoing-like control algorithm to keep one section of the net in view and centered at a time and to recognise when the vehicle has reached the bottom of the cage. A dataset was collected in real-life conditions at the CROMARIS fish farm near Ugljan, Croatia and will serve as a basis for training the image processing algorithm providing feedback for the controller. Finally, the developed basic image acquisition and control interfaces will be tested on the Blueye vehicle.