Search

This article presents an online ROS-integrated YOLOv8-based sea urchin detector for Autonomous Underwater Vehicles (AUVs). In order to validate the performance of the system, a realistic simulation environment has been generated using the Stonefish simulator. The simulation setup, including the vehicle and the environment, recreates both the sensors and actuators of the AUV explorer in order to be able to run the detection algorithm under simulation., Peer Reviewed

This paper presents preliminary data and results from field experiments in which a stereo-camera-unit was deployed at approximately 100m depth, capturing images of sea fauna. The stereo vision system is described, as well as the image processing techniques used to enhance brightness and facilitate fish detection., Peer Reviewed

In addition to the potential global impact of climate change on marine ecosystems, the extensive use of high-impact fishing methods is a primary catalyst for benthic biodiversity degradation in the Mediterranean Sea. Implementing fishery no-take zones (FNTZs) emerges as a key measure for the sustainable recovery and management of overexploited stocks and habitats. To identify appropriate geographical scales for their implementation, it is crucial to understand the spatial connectivity of species and ecosystem functioning during long periods. Therefore, it is necessary to implement robust spatio-temporal multiparametric monitoring procedures, allowing the synchronous collection of biological (i.e., image-based), oceanographic and geochemical data. For this, we developed a spatial cooperative network of fixed (i.e., landers) and docked mobile (i.e., AUVs) platforms with wireless intercommunication capability (i.e., by acoustic modems). This system is designed for intelligent observation monitoring and mapping (i.e., AI-based recognition of species and bioturbation features) over extended periods with real-time, remote supervision and data transmission through the water column to an ASV., Peer Reviewed