Predefined-time formation tracking for AAVs with prescribed performance: A neural network-based reinforcement learning approach

Abstract

This paper proposes a solution to the predefined-time formation tracking problem of autonomous aerial vehicles (AAVs). To enhance the systems’ coordination performance and convergence rate, a predefined-time hierarchical control (PTHC) framework is developed. Furthermore, a prescribed-performance control mechanism is introduced to impose prior constraints on both transient and steady-state behaviors during the design stage, thereby ensuring that the tracking error remains within predefined bounds. For the state estimation process, a predefined-time distributed estimator based on nonsingular sliding mode control is designed, effectively avoiding the singularity issues commonly encountered in conventional sliding mode approaches. In the control implementation, an adaptive strategy integrating neural networks with reinforcement learning is proposed to improve the systems’ resistance to external disturbances, actuator faults, and modeling uncertainties. This combination significantly enhances the robustness and reliability of the controller, ensuring stable operation under complex and dynamic environments. The designed control strategy ensures the realization of the desired formation tracking goal within the predefined time. The paper further establishes a set of sufficient criteria to ensure predefined-time stability. Finally, extensive simulation studies are conducted to verify the proposed control methods’ efficiency, robustness, and practical applicability.