IJCAS

A time-delayed control (TDC) method is known as a simple, robust and non model-based control scheme that requires the fast sampling time, the accurate measurement of joint acceleration signals, and the accuracy of the inertia model of a robot manipulator. Among them, sampling time and acceleration signals are hardware dependent and can be solved. Then a user specified inertia model becomes a key role for the performance of TDC. When the selection of the diagonal element of the inertia matrix of a robot manipulator is used, the ill selection of the constant inertia matrix may lead to the poor tracking performance as well as instability. In addition, an appropriate selection of an inertia matrix for different tasks of the robot is not easy. Therefore, in this paper, an intelligent way of using a neural network is proposed to compensate for the deviation of the constant inertia matrix of a robot manipulator. The role of the neural network is to improve the tracking performance of a robot manipulator by compensating for the deviated error of the inertia matrix while satisfying the stability bound. Simulation studies of a three link robot are presented to confirm the proposal.

Keywords: Inertia compensation, neural network, robot manipulators, stability, time-delayed control