Neural-Enhanced Sliding-Mode Control for Quadrotor Helicopters Operating in Dynamic and Uncertain Environments

Robust altitude tracking and attitude stabilization of a quadrotor helicopter (QH) operating in dynamic and uncertain environments remain challenging due to unknown external disturbances, time-varying parametric uncertainties, and practical actuator and sensor constraints. To address these challenges, this article proposes a neural-enhanced sliding-mode control (NE-SMC) framework that integrates a lightweight multilayer perceptron neural network (MLP-NN) with an SMC backbone within a unified adaptive control architecture. The neural module employs a sliding-mode-inspired weight-update law to enable real-time compensation for disturbances and uncertainties without requiring explicit disturbance models or prior knowledge of system parameters, thereby strengthening system-level robustness and adaptability. A rigorous Lyapunov-based analysis guarantees closed-loop stability and finite-time convergence of the tracking errors. Extensive high-fidelity simulations, including deterministic scenarios and a 100-run Monte Carlo study, are conducted under severe wind disturbances and turbulent conditions, up to 25% mass and inertia variations, actuator saturation, and sensor constraints. Comparative results demonstrate that the proposed NE-SMC achieves substantially improved tracking accuracy and robustness compared with standard SMC and conventional hybrid SMC (H-SMC) schemes. Specifically, the root-mean-square error (RMSE) is reduced by 83.3%–88.9% relative to standard SMC and by 75%–87.5% relative to conventional H-SMC. These results confirm that the proposed framework provides a robust, constraint-aware, and computationally efficient control solution suitable for real-time QH applications, aligning with the requirements of intelligent systems in complex, uncertain environments.