Abstract

Marine ecosystems face unprecedented threats from pollution, climate change, and human activities. This paper presents the design and implementation of an autonomous underwater robot equipped with AI-powered computer vision for real-time detection and classification of marine debris, including ghost nets, plastic waste, and microplastic concentrations. The robot employs a YOLOv8-based detection system trained on a custom dataset of 15,000 underwater images, achieving a mean average precision (mAP) of 0.89 for ghost net detection. Integrated with oceanographic sensors, the system simultaneously monitors water quality parameters including temperature, pH, dissolved oxygen, and turbidity. Field tests conducted in the Mediterranean Sea (Italian and Turkish coastlines) demonstrated the system's capability to survey 2.5 hectares per hour at depths up to 50 meters. The collected data feeds into a cloud-based dashboard providing researchers and environmental agencies with real-time ecosystem health assessments.