Decentralized Model-Free Monitoring of Multi-UAV-Multi-USV Systems Using Sparse Data and Bayesian Learning

Abstract

Although significant progress has been made in coordinating multi-unmanned surface vehicle (multi-USV or USVs) systems over the past decades, consistent monitoring (or tracking) of such systems remains challenging as they do not share data with monitoring systems, further exacerbated by observed data inaccuracies and sparsity. To tackle the complicated issue, we hereby introduce the multi-unmanned aerial vehicle (multi-UAV or UAVs) system to monitor the multi-USV system. Therein, by introducing a sparse-Bayesian-learning-based (SBL-based) algorithm, the multi-UAV system can identify the potential coordinated dynamics of multi-USV system via only noisy and limited data. Then, by employing the Kalman filter (KF), the proposed approach can predict and update real-time data and optimize trajectory estimation for USVs, and enhance coordination control in the multi-UAV system to achieve coordinated monitoring. Finally, comparative simulations against the traditional control method, conducted under varying noise levels and data availability ratios, demonstrate the effectiveness and superiority of the proposed method.