Trajectory Tracking Prediction of Multiple-UAVs Formation Based on Gated Cyclic Convolution Neural Network

Abstract

Unmanned Aerial Vehicles (UAVs) represent a typical example of underactuated and strongly coupled nonlinear systems, renowned for their high speed, maneuverability, and endurance, making them a prominent research focus in the fields of national defense and air defense. However, controlling a fixed-wing UAV poses challenges due to its susceptibility to external interference and the inherent complexity of the flight environment. To address these challenges, this study adopts a novel approach based on the gated cyclic convolutional neural network (GCCNN) architecture. By leveraging the unique structure of GCCNN, this research successfully solves the four control input signals of a fixed-wing UAV and employs the gated convolutional neural network for trajectory control and prediction. The utilization of cyclic convolution offers distinct advantages, enhancing the accuracy of UAV trajectory prediction and improving the overall trajectory prediction effectiveness.