The AQUASTAR project targets combined autonomous inspection and repair of holes in aquaculture fish cage nets . Holes in the net represent the main reason for fish escape and is a challenge in aquaculture fish farming in traditional net-based net pens.
To reduce the risk of escapes i n Norway, recent regulations state that inspections must take place before, during, and after operations that involve the net . This has led to more frequent net inspections, but closer investigations have revealed several occurrences where both small holes and large tears have gone undocumented. This creates a need for enhanced inspections and, most preferably, immediate repair of holes.
Today, net inspections are carried out manually and include a remotely operated vehicle (ROV) and an ROV pilot. However, manually inspecting repeating net patterns for hours on end may be tedious and can increase the risk of missing holes in the net. Detected h oles are often reported by denoting the hole’s approximate size, depth, and compass heading. Next, a report stating needs for repairs is issued to companies that perform hole repair using divers . However, the time between a hole is detected and the repair is done may range between hours to days, leaving a time window for holes to grow in size and for fish to escape. Moreover, while it may be expensive to hire divers for hole repair, divers are exposed to risks that may lead to both permanent injuries and even death . This strongly advocates for both enhanced hole detection systems and alternative net repair solutions.
The AQUASTAR project will develop solutions for ROVs with manipulators enabling automatic hole localization , automatic vehicle positioning relative to holes, and hole repair . This will serve as a tool for ROV pilots to ac hieve enhanced net inspections and the opportunity to immediately repair holes in the net , both manually and autonomously . The project will build up on previous developments within net- and object-relative navigation systems and vehicle-manipulator operations. Moreover, AQUASTAR aims to develop and continuously update models for detecting holes in the net, and to publish open- source datasets for hole detection. The models and datasets will include several varying conditions present in the aquaculture industry such as illumination conditions based on time of day, season and weather, varying particles in the water, varying levels of biofouling, and more .
The AQUASTAR project is a n industrial research and collaboration project between Njord Aqua, Måsøval, Sintef Ocean, and NTNU, and is supported by the Research Council of Norway.