DESIGN AND IMPLEMENTATION OF AUTONOMOUS UNDERWATER VEHICLES FOR SUBSEA PIPELINE INSPECTION

This project presents the design and deployment of an Autonomous Underwater Vehicle (AUV) for subsea pipeline inspection. Recognizing the vulnerabilities inherent in offshore pipelines ranging from corrosion and material fatigue to damage by external factors. The study emphasizes the creation of a cost-effective, vision-driven inspection system. Unlike conventional inspection methods that rely on complex and expensive sonar arrays or Doppler Velocity Logs (DVLs), this work introduces a streamlined solution based purely on optical sensors and machine learning algorithms. Leveraging a YOLOv8-based detection framework and Visual Simultaneous Localization and Mapping (Visual SLAM), the AUV demonstrated reliable performance in segmenting pipeline sections and identifying structural anomalies without auxiliary navigation aids. Experimental evaluations in controlled environments showed a high mean average precision for object detection, robust obstacle avoidance and navigation capabilities. Although real-world subsea challenges such as turbidity and low- light conditions were beyond the scope of this initial testing phase, this system offers a promising, low- cost alternative for future subsea monitoring operations, aligning with industry needs for accessible and sustainable offshore asset management.