Adaptive Neural‐Augmented High‐Order Sliding Mode Control for Integrated UAV Guidance in Uncertain Environments

This study proposes a novel integrated guidance and control (IGC) framework for unmanned aerial vehicles (UAVs) operating in uncertain and time‐varying environments. The core of the framework is an Adaptive Continuous High‐Order Sliding Mode Controller (ACHOSMC) augmented with a fully online neural network (NN) that adaptively tunes control gains in real time based solely on instantaneous tracking errors eliminating the need for any offline training or prior model knowledge. Unlike conventional sliding mode schemes, the continuous high‐order formulation significantly suppresses chattering while ensuring finite‐time convergence, even under severe disturbances and modeling uncertainties. The nonlinear UAV dynamic model, decomposed into independently modeled pitch, yaw, and roll channels with structured interchannel uncertainties, provides a rigorous benchmark for evaluation. Computational feasibility is experimentally validated on an STM32‐class embedded processor, demonstrating a total adaptation‐and‐control cycle time of 0.6 ms, enabling update rates well above 100 Hz for agile maneuvering. Comprehensive simulations compare the proposed ACHOSMC–NN with PID, SMC, and CHOSMC controllers in demanding trajectory‐tracking scenarios. Results show consistent improvements in tracking precision, convergence speed, robustness to disturbances, and smoothness of control signals. These contributions position the proposed method as a computationally efficient, chattering‐free, and adaptively self‐tuning IGC solution for UAV missions such as autonomous logistics, precision agriculture, infrastructure monitoring, and high‐maneuverability inspection tasks.