Adaptive Neural Network-Based Model Path-Following Contouring Control for Quadrotor Under Diversely Uncertain Disturbances

Quadrotors, while versatile, are vulnerable to unpredictable environmental disturbances, including turbulence, gusts, and ground effects, making precise path-following control a formidable challenge. This letter introduces an adaptive neural network-based predictive control framework tailored for quadrotors. This framework synergistically integrates a high-level Model Path-Following Contouring Control (MPFCC) with a low-level Adaptive Neural Network-based Feedback Linearization Controller (ANN-FBLC) to ensure adaptive path-following amidst uncertain disturbances. The ANN-FBLC, at its core, leverages neural networks for real-time disturbance approximation, establishes an integrator model via feedback linearization, and meticulously designs adaptive network weights, ensuring precise reference state tracking. Meanwhile, the high-level MPFCC uses the linear system model obtained by feedback linearization to optimize the path reference target, achieve proactivity against transient abrupt perturbations and provide reference states and control for the low-level ANN-FBLC. Experimental evaluations of the efficacy of our proposed control strategy, particularly in guiding a quadrotor's path-following amidst diverse disturbances.