ANN-MPC: An Adaptive Neural Network-Based Nonlinear Model Predictive Control Method for Robust Trajectory Tracking of Quadrotors

Quadrotor systems, despite their extensive deployment in diverse application domains, face substantial performance degradation due to three predominant factors: unmodeled aerodynamic phenomena, persistent wind disturbances, and unanticipated payload variations. These challenges collectively compromise precise trajectory tracking in dynamic environments. To address these limitations, this study introduces an integrated control architecture combining 1) an adaptive neural network disturbance observer (ANNDO) with guaranteed Lyapunov stability through carefully designed adaptive laws, and 2) a modified model predictive control (MPC) scheme enhanced with real-time disturbance compensation. The proposed framework demonstrates substantial performance improvements, achieving over 80% tracking accuracy enhancement in numerical simulations during agile flight, and over 50% improvement in physical experiments involving complex disturbance rejection, compared with baseline MPC.