Teleoperation Control of a Position-Based Impedance Force Controlled Mobile Robot by Neural Network Learning: Experimental Studies

Abstract

Effective haptic performance in teleoperation control systems can be achieved by solving two major problems: the time-delay in communication channels and the transparency of force control. The time-delay in communication channels causes poor performance and even instability in a system. The transparency of force feedback is important for an operator to improve the performance of a given task. This article suggests a possible solution for these two problems through the implementation of a teleoperation control system between the master haptic device and the slave mobile robot. Regulation of the contact force in the slave mobile robot is achieved by introducing a position-based impedance force control scheme in the slave robot. The time-delay problem is addressed by forming a Smith predictor configuration in the teleoperation control environment. The configuration of the Smith predictor structure takes the time-delay term out of the characteristic equation in order to make the system stable when the system model is given a priori. Since the Smith predictor is formulated from exact linear modeling, a neural network is employed to identify and model the slave robot system as a nonlinear model estimator. Simulation studies of several control schemes are performed. Experimental studies are conducted to verify the performance of the proposed control scheme by regulating the contact force of a mobile robot through the master haptic device.