A sea cage can hold up to 200,000 farmed salmon, and if the cage sustains damage, the fish could swim out through the opening and escape, which is a scenario that the salmon aquaculture industry wants to avoid, according to the Norwegian University of Science & Technology (NTNU).

Not only do escapes lead to large losses for the industry, but the escaped farm-raised salmon could also mix and interbreed with wild populations, NTNU said.

Monitoring life in the cages is important for other reasons as well, NTNU said, such as ensuring good fish welfare: What is the health condition of the fish? How serious is the salmon lice problem? Do the cages need to be cleaned?

Human divers and underwater vehicles controlled by operators on land are commonly used to check the conditions in sea cages, but both methods can disrupt and stress the fish and limit the frequency of inspections.

Robotics and biology researchers have been trying to find out which monitoring methods disturb fish least, NTNU said, noting that tests with a robotic turtle swimming around the cage to film the equipment and fish have been able to do the inspection job better and more gently.

The experiments show that the fish are only negligibly scared or stressed by the robotic turtle. They swim calmly and fairly close to the turtle, whereas they keep away from the intruders in experiments with divers and thruster-driven underwater robots, NTNU said.

"The overall purpose of the experiments wasn't just to test the turtle robot but also to investigate what characteristics robots being used in the aquaculture industry should have," said Maarja Kruusmaa, a professor with the NTNU department of engineering cybernetics in Norway as well as the Tallinn University of Technology in Estonia. "We've found that the most crucial characteristics of the surveillance robot are its size and speed, whereas color and motor noise hardly matter at all."

The turtle robot's small size and slow movements are the characteristics that make it less disturbing to the fish, NTNU said, adding that the fact that it resembles an organism that lives in the ocean is less important.

"The conclusion turned out to be the opposite of our expectations. The fact that the robot looks like a marine animal doesn't seem to play any role at all, and that's actually good news: It means we don't have to build the robots to be fish- or turtle-like. That will make it cheaper to develop and use robots in this new field of application to monitor marine organisms," Kruusmaa said.

The research indicates which factors are important when developing robots for the fish farming industry or for monitoring fish in their natural setting, NTNU said.

Kruusmaa and Jo Arve Alfredsen, an associate professor in the NTNU department of engineering cybernetics, published an article about their findings in Royal Society Open Science.

Kruusmaa and Alfredsen are both employed by NTNU AMOS -- the Center for Autonomous Marine Operations & Systems. AMOS is developing new types of underwater vehicles and new offshore monitoring methods as their focus areas.

Robots like the robotic turtle can provide fish breeders with online updates and monitoring of life in the sea cage. The turtle can also be connected to various measuring instruments and sensors.

Using robotic technology instead of divers for surveillance allows monitoring to continue without interruption, NTNU said, noting that this continuity can contribute to quicker responses, greater predictability, better fish welfare and lower mortality.

The turtle robot, named U-CAT, was developed at Tallinn University of Technology and was originally designed for underwater archaeology applications. The idea was to use it to investigate shipwrecks on the sea floor, so it was designed as a small and very maneuverable robot.

Alfredsen realized that the robot could be used in aquaculture because it had precisely these properties.

The experiments in the sea cages at SalMar have shown that this robotic technology can also benefit the aquaculture industry.