Abstract
The ship's event-triggered tracking algorithm, which combines game theory and remote sensing. The technique addresses issues with conventional continuous and discrete control techniques, which consume excessive amounts of energy and prematurely wear out actuators due to frequent updates. In order to locate ships, this paper first employs the YOLOv8 deep neural network and radar-based remote sensing. After determining the distances and connections between ships, the target trajectory is altered using a game-theoretic framework. In order to reduce needless computations and communications, the controller only activates when significant events take place. This is not the same as conventional time-triggered techniques. Good results were obtained from tests on two benchmark datasets: the Synthetic Aperture Radar Ship Dataset (SAR-Ship-Dataset) and the SAR Ship Detection Dataset (SSDD). On SSDD, the proposed algorithm achieved 92% accuracy and 93% sensitivity, while on SAR-Ship-Dataset, it achieved 93% accuracy and 91% sensitivity. These results lay the foundation for the development of more efficient marine tracking systems by demonstrating that event-triggered tracking can maintain high tracking accuracy while using less energy.
Keywords
DeepSORT
game theory
Marine Surface Vehicles
Multi-Object Tracking remote sensing YOLOv8