Global Predefined-Time Neural Network-Based Performance Control for UAV Safe Formation With Input Saturation and Disturbance Estimation

Abstract

This study presents a global predefined-time adaptive control method with prescribed performance for quadrotor UAV safe formation subject to external disturbances. The dynamic models of quadrotor UAVs are first transformed into position and attitude subsystems. Then, a global neural network-robust control strategy with switching functions is proposed by introducing an $n$ -th-order continuously differentiable function. On this basis, a global predefined time-prescribed performance controller is developed by utilizing a performance function, which ensures both position and attitude subsystems converge globally to specified accuracy within a given time. During flight, obstacle effects are also considered, and an obstacle avoidance strategy is designed using Barrier Lyapunov Functions (BLFs) to guarantee the safety for the UAV formation. Furthermore, disturbance estimation and input saturation are incorporated into the controller scheme, which allows the UAV to compensate more accurately for external perturbations and better meet real-world engineering requirements. Finally, the effectiveness of the proposed algorithm is verified through comprehensive numerical simulations and real-world flight experiments. Note to Practitioners—This work presents a robust control framework for multi-UAV formation flight in complex environments, designed to achieve fast, precise, and collision-resistant operation under external disturbances such as wind gusts and aerodynamic effects. The method enables a fleet of UAVs to accurately converge to the desired formation within a strictly predefined time and automatically avoid both static and moving obstacles. Furthermore, it explicitly accounts for critical real-world constraints including input saturation and unknown disturbances, significantly enhancing its practical applicability and deployment potential. Comprehensive simulations demonstrate that the proposed scheme offers high reliability, excellent tracking performance, and effective disturbance rejection under challenging operational conditions.