ESO-Based Adaptive Neural Network Control for Quadrotors Under Multiple Uncertainties

This paper introduces a robust framework for mitigating internal and external disturbances in quadrotor systems. Specifically, a radial basis function neural network (RBF-NN) is utilized for the estimation of model uncertainties, while extended state observer (ESO) compensates for external disturbances and RBF-NN approximation errors. This dual estimation mechanism provides stronger theoretical guarantees, enhancing the system's capacity to handle complex scenarios. Additionally, a Lyapunov-based adaptive control strategy is employed to dynamically adjust control gains, managing variations in thrust and torque coefficients due to rotor dynamics. The incorporation of a projection operator in the parameter update law ensures the boundedness of parameter estimates. The proposed method demonstrates improved control accuracy and stability under multiple unmodeled uncertainties.