Field experiments on real-time autonomous marine surveillance

Monitoring and protecting critical marine infrastructures–such as offshore wind farms, data cables, and subsea pipelines–demands reliable, autonomous surface vessels capable of operating efficiently in real-world conditions. In these scenarios, the area of interest is typically known in advance, and the primary objective is to detect intruders or any other anomaly. This paper introduces a comprehensive Guidance, Navigation, and Control (GNC) architecture for Marine Autonomous Surface Ships (MASS), with particular emphasis on model-scale validation conducted during a relevant environmental testing campaign. The underlying hypotheses and methodologies are detailed and thoroughly discussed. Building on previous simulation-based studies that validated individual components, this work focuses on verifying the integration of the stochastic coverage planner, the tracking controller, and the state estimation subsystem through simulation and real-world field trials. The experimental results obtained from tests conducted at an outdoor site demonstrate successful completion of surveillance missions, real-time compliance, and the system's adaptability to dynamic changes and environmental disturbances. These findings confirm the feasibility of transferring complex autonomous behaviours from simulation to operational platforms. The proposed framework provides solid validation for future research and development of the architecture, enabling integration with the perception system and collision-avoidance capabilities to detect and avoid obstacles and intruders.