A network of dynamic assets equipped with surveillance sensors (e.g. radar) operates to keep the vessel traffic in a given region of interest (ROI) under control. The problem addressed in this work is the coordination of the network by providing an optimal routing plan with respect to given objectives and given operational constraints for a temporal horizon of few days. Specifically, for the sea counter piracy scenario, this work proposes to improve the vessel traffic surveillance by redirecting the moving assets in those areas of the ROI where there is a lack of information (e.g. lack of satellite AIS coverage) and at the same time favorable METOC conditions for piracy activity.
To this purpose, a surveillance risk metric is defined and minimized with respect to asset waypoints and speed. This metric is defined by combining satellite AIS performance predictions and piracy activity group (PAG) maps, predicted by fusing METOC forecasts, together with asset sensor performance. A centralized hierarchical fusion architecture is taken into account in defining the metric. Additional conflicting objectives are also taken into account including asset mission costs and network spatial coverage. The problem is constrained by the asset kinematic and operational limits (such as asset endurance). The optimization problem can be also refined by including spatial and temporal constraints such as denied areas and temporal windows in which an asset is available (this capability is not implemented yet). Given the asset initial states (position, velocity and heading), and the constraints, the optimization provides the optimal set of way points and the speed between consecutive waypoints for the navigation of each asset.
The optimization is carried out by using a state of the art evolutionary multi-objective algorithm. The optimizer simultaneously minimizes (or maximizes) conflicting objectives by implementing the concept of Pareto dominance. The algorithm is able to provide a set of solutions which are distributed on the so called Pareto frontier in the multi-dimensional objective space. These solutions represent the optimal trade-offs among the selected objectives; the user of the system can explore the frontier and choose the solution that best fits his/her preferences depending also on the situation at hand.