Neural Network Adaptive Tracking Control for Continuum Robots Considering Modeling Uncertainties

Abstract

Continuum robots are widely used in rescue, medical, and other special operating environments due to their excellent structural characteristics. In this paper, we use Euler-Bernoulli beam equation to establish the dynamical model of a linear driven continuum robot based on the constant curvature hypothesis. Since there always exist unknown disturbances or system uncertainties in the modeling and control process of an continuum robot, the robust controller design for such system is of great important in practical applications. To solve this problem, we propose an adaptive compensated tracking controller based on an improved Radial Basis Function (RBF) Neural Network (NN) to compensate modeling uncertainties and unknown external disturbances, and to achieve better tracking performance of the closed-loop system state variables. Stability analysis proves that the system is stable and the uniform boundedness of states is guaranteed. Finally, Numerical simulation results verify the efficient performance of the designed NN-based adaptive tracking controller.