Design of Adaptive RBFNN and Computed-torque Control for Manipulator Joint Considering Friction Modeling

In this paper, we aim to improve the tracking performance of the manipulator joint system by establishing accurate friction model based on the Stribeck model and the cubic polynomial method. Meanwhile, in view of the established system model, an adaptive Radial Basis Function Neural Network (RBFNN) compensation computed-torque controller is designed for the manipulator joint system. Firstly, we consider the friction modeling process at low- and high- velocity regions to advance the model accuracy, and identify the parameters in the friction model equation offline via the particle swarm optimization (PSO) algorithm. Secondly, an adaptive RBFNN algorithm is developed to analyze the unmodeled dynamics online and introduce it to the computed-torque controller design. After that, we further conduct the stability analysis for the proposed controller based on the Lyapunov stability criterion. Finally, the self-developed manipulator joint platform introduction, the simulation experiment and the contradistinctive experiments are given to illustrate the effectiveness of designed controller.