Learning-based Finite-time Distributed Formation Control for Fixed-wing Unmanned Aerial Vehicles

This paper studies the learning-based finite-time distributed formation control problem for a group of fixed-wing unmanned aerial vehicles (UAVs). Using position information obtained from sensors, a novel learning-based finite-time distributed controller with a saturation function is developed. Firstly, a six-degree-of-freedom fixed-wing UAV model is established and transformed into a double-integrator model using linear feedback linearization. For practical implementations, the velocities and overloads of the controlled fixed-wing UAVs are constrained within desirable limits by injecting appropriate saturations into the loops. The finite-time formation control objective is demonstrated if the connectivity topology meets the spanning tree criterion. The Lyapunov theory is used as a fundamental tool to guarantee the stability of the closed-loop system. Furthermore, an improved learning-based finite-time distributed control scheme is proposed to obtain the approximately optimal control laws for the error systems. Finally, two numerical examples are employed to verify the theoretical results.