Adaptive Neuro-Fuzzy Sliding Mode Tracking for Quadrotor UAVs

Quadrotor systems offer significant potential, yet precise trajectory tracking remains challenging due to their complex dynamics and nonlinear characteristics. In particular, the unmodeled dynamics of the translational subsystem and the nonlinearity of the rotational subsystem present substantial obstacles. This study proposes a dual-loop structure integrating neural networks (NN) and fuzzy control techniques. By combining the sliding-mode control (SMC) method with adaptive laws, the approach achieves a fast tracking response, improved transient performance, and reduced chattering. An adaptive neural control based on a switching surface addresses the unmodeled dynamics of the outer loop and performs position control. Meanwhile, a dissipative fuzzy sliding-mode controller for the inner loop manages nonlinearity and input uncertainty, ensuring robust attitude control. The designed controllers’ stability was verified using Lyapunov theory, and the simulation of various trajectories demonstrated the robustness and potential of the proposed dual-loop neuro-fuzzy tracking controller. Note to Practitioners—In many real-world applications, PID controllers are often preferred over model-based controllers due to the significant nonlinearity and unmodeled dynamics of quadrotors. To address these challenges and ensure robustness against disturbances, we developed a novel controller that integrates SMC, NN, fuzzy logic, and dissipative performance. Unlike PID controllers, which require extensive tuning and optimization for specific tasks, the adaptive and modeling capabilities of fuzzy logic and NN enable stable flight without additional adjustments. This approach offers a cost-effective solution for managing quadrotors in applications such as search and rescue, transportation, and other tasks involving complex trajectories, unknown dynamics, and external disturbances.