Trajectory Tracking of a WMR Using a Neuro-Adaptive MRAC Control With Dynamic Uncertainties

This work presents a Neuro-Adaptive Model Reference Adaptive Control (NA-MRAC) strategy for trajectory tracking in wheeled mobile robots (WMRs) subject to nonlinear dynamics, load-dependent variations, and unmodeled uncertainties. The proposed approach integrates an inverse-kinematics controller with a dynamic NA-MRAC structure that merges a radial-basis-function neural network and a regressor-based adaptive mechanism, enabling simultaneous compensation of parametric deviations and nonlinear residual dynamics. A complete stability analysis of the cascade kinematic–dynamic architecture is provided, establishing bounded-error behavior under the derived adaptation laws. The method is assessed in simulation using a Pioneer 3DX platform in MobileSim, including scenarios with sudden payload changes, nonsmooth trajectories with directional discontinuities, and a comparative evaluation against a non-adaptive MRAC implementation and a regression-based MRAC controller. The results show that the NA-MRAC provides improved robustness and disturbance rejection, with faster transient response and reduced tracking error in the evaluated scenarios, compared to the other approaches.