Adaptive Hybrid Force/Position Control Using Neuro-Adaptive Observer for Dual-Arm Robot

Abstract

The paper has developed an adaptive hybrid force/position control scheme using neural network observer for controlling dual-arm robotic system considering system uncertainties. Firstly, an overall dynamics of the system including the manipulators and the object are derived using Euler-Lagrangian principle. Then, a neural-adaptive observer is designed to estimate the object velocities using a radial basis neural network. Based on the observed velocities, an adaptive hybrid force/position controller is proposed to compensate dynamic uncertainties without force measurement. The neuro-adaptive observer and the controller learning algorithms have been derived according to Lyapunov stability principle in order to guarantee asymptotical convergence of the closed loop system. Finally, simulation work has been carried out to validate the accuracy and the effectiveness of the proposed approach.