An Adaptive Interception and Tracking Strategy of Unmanned Surface Vehicles Based on Multi-Model Prediction

This paper presents a novel adaptive interception and tracking strategy for unmanned surface vehicles (USVs). This strategy provides a flexible approach by considering the target’s state and the performance of the interceptor, offering enhanced adaptability compared to traditional approaches. The proposed multi-model prediction framework integrates a Kalman filter and a long short-term memory (LSTM) model. This design enables the prediction module to achieve effective forecasting performance in both linear and nonlinear motion scenarios. The hybrid control strategy combines two controllers to provide robust performance and adaptability in dynamic environments. The designed backstepping controller stabilizes the system by adjusting the control inputs to match the target’s trajectory. The designed sliding-mode controller ensures successful interception by swiftly responding to changes in the target’s path. Case studies demonstrate the strategy’s improved accuracy and reliability, achieving significant performance gains across various scenarios.