Nonlinear NN-Based Perturbation Estimator Designs for Disturbed Unmanned Systems

This paper addresses the challenge of estimating perturbations in a classical unmanned system caused by a combination of internal uncertainties within the system and external disturbances. To accurately approximate these hard-to-measure perturbations, a novel nonlinear radial basis function neural network (RBFNN)-based estimator is introduced. This estimator is designed to reconstruct the perturbation structure effectively. The study demonstrates that utilizing RBFNN-based estimator designs, coupled with Lyapunov stability analysis, leads to achieving asymptotic estimation results. The effectiveness of the proposed perturbation estimation approach is validated through simulations conducted on both an unmanned marine system and a quadrotor system.