Autonomous Aerial-Aquatic Robot Swarm for Offshore Oil Pipeline Inspection: Leak Detection Sensitivity, Defect Classification Accuracy, and Swarm Coordination Protocol Validation

This study investigates autonomous aerial-aquatic robot swarm for offshore oil pipeline inspection, testing leak detection sensitivity, defect classification accuracy, and multi-agent coordination protocols on a 48-km test pipeline segment within the context of marine robotics and offshore infrastructure inspection, an area of growing scientific importance given its implications for offshore pipeline integrity management regulation compliance, predictive maintenance scheduling, and autonomous marine inspection robot certification framework. Using controlled leak trials (6 defect types at 8 flow rates), live field inspection trial on decommissioned 48-km Chevron test pipeline, swarm coordination latency measurement, and defect classification by onboard CNN vs. human expert comparison, we examine distributed task allocation protocol (auction-based, 200ms round-trip latency) enabling aerial UAVs to flag surface methane anomalies for prioritized AUV inspection, reducing redundant coverage and increasing per-km inspection rate vs. single-platform approach in 48 controlled leak trials (6 defect types x 8 flow rates); 8 full swarm field deployments on 48-km pipeline; 384 pipeline inspection images classified by swarm CNN and 3 human experts drawn from 48-km decommissioned Chevron pipeline in Gulf of Mexico shallow shelf (10-40m depth), with topside coordination vessel and satellite relay for swarm communication; controlled leak trials at GCEI marine test facility (Barataria Bay). Results indicate that swarm achieves 0.84 L/min minimum detectable leak flow (94.2% sensitivity at 1 L/min threshold), 4.2% false alarm rate; CNN defect classification 91.8% accuracy vs. expert 94.2%; swarm completes 48-km inspection in 6.4h vs. 28h single AUV (4.4x speedup) (p < 0.001), with 94.2% sensitivity at 1 L/min; 91.8% CNN accuracy; 4.4x inspection speedup as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to marine robotics and offshore infrastructure inspection and carry actionable implications for the design of programs and policies targeting offshore pipeline integrity management regulation compliance, predictive maintenance scheduling, and autonomous marine inspection robot certification framework.