AI‑driven maritime surveillance: Moving‑target routing optimization

Monitoring vessels across vast maritime areas presents a unique challenge for authorities. While data systems are powerful, they often fall short when trying to track non-characterized vessels in real-time. That is why aerial reconnaissance has become essential, yet deploying these assets efficiently is far from simple. In the DIGEST project, we tackled the specific difficulty of coordinating manned and unmanned aircraft to monitor moving targets, all while working within strict flight endurance limits. Our goal was to move beyond static planning and create a system that truly understands the dynamic nature of the sea.

To solve this, we introduced a new mathematical framework called the Team Orienteering Problem with Moving Targets (TOP-MV). Unlike traditional routing problems where destinations are fixed, our model accounts for vessels that are constantly changing position. By breaking time down into manageable windows, we converted this puzzle into a solvable optimization problem. This approach allows us to plan routes for a mixed fleet of drones and manned aircraft simultaneously, ensuring that every asset is used where it can provide the most value without running out of fuel.

Of course, no model is perfect. Our current approach assumes we know the vessel paths at the start of the mission and plans a single flight per asset. While this works well for standard reconnaissance missions where aircraft can refuel between sorties, future versions of our work will aim to handle more dynamic situations, such as sudden changes in vessel direction or the need for multiple flights in a single day. We also recognize that scaling up to massive fleets will require new computational strategies to keep things running smoothly.

Despite these areas for growth, our results confirm that this approach offers a solid foundation for smarter maritime surveillance. It proves that we can significantly improve how we allocate resources and cover large areas without needing minute-by-minute precision. By establishing this new baseline for routing moving targets, we hope to pave the way for decision-support systems that make coastal security and environmental monitoring more effective and responsive.

We are excited to see how this research evolves and how it can be integrated into real-world operations. The full details of our methodology and findings are now published in Computers & Industrial Engineering, where we invite fellow researchers and industry professionals to explore the potential of AI-driven fleet coordination.