Teleoperation System, Neural Network Control

Time delay exists in teleoperation systems due to long transmission of signals. Space robots have link flexibility due to its' light weight and fast speed. All these make control of teleoperation systems difficult. This paper aims at controller design of rigid robots and flexible robots and controller design of teleoperation systems using neural network identification and Smith predictor control methods.Dynamic neural network is used to approximate the dynamics nonlinear function and linearize the nonlinear dynamics model for rigid robots. Smith predictor is used to eliminate the influence of time delay and the rigid teleoperation system is designed. The elastic deformation is analyzed using the assumed-mode method for flexible robots and the dynamics equation of a single flexible manipulator is established using Lagrange method. The singular perturbation approach is used to decompose the robot system into a slow subsystem and a fast subsystem. Special controller is designed for each subsystem. Smith predictor is used to design the flexible teleoperation system. On-line neural network weight tuning is used.