Scientists have created an artificial intelligence-based system to prevent fish from escaping from salmon farms. This has been reported by the Norwegian University of Science and Technology.
Frequent checking of marine cage nets suspended underwater aims to help prevent fish from escaping. Escaped farmed salmon can carry diseases and can also migrate upstream with wild fish, interrupting the spawning process. These factors only weaken wild salmon populations, so escapes should be avoided by all means. Networks are currently being controlled by remote-controlled underwater vehicles.
Norwegian scientists decided to automate this process. Even navigating through the many meshes is difficult, as the car should get close enough to notice the gap, but not get tangled up in the grille. In addition, the robot must accurately determine its position, remain stable relative to the base, and resist currents.
In addition, the program must be able to distinguish all networks from damaged ones. To do this, the neural network was trained on a large image sample of all meshes to generate an image of the “correct” grid. If an anomaly is detected in the appearance, the program gives a signal and people decide to make a repair.
The authors hope their robot will enable salmon producers to meet the Norwegian government’s new cage reliability requirements.