Control of a multiple link robot arm at very high speeds for an industrial application

Abstract

Control of a nonlinear multi-link golf/tennis swing robot manipulator at very high speeds of 100 miles an hour while following the desired trajectory accurately, is a very difficult and challenging task. In addition, the load carried by the end-effector with each trajectory changes due to different size golf clubs and tennis rackets. Furthermore, the human and manipulator interactive forces, which vary with humans, are complex and they induce disturbances at the end effector. Therefore, a robust CMAC neural network (NN) controller is introduced to control the multiple link robot arm as the nonlinear dynamics of the arm, the mass of the golf clubs/tennis rackets and the human interactive forces are not known beforehand accurately for this very high-speed application. Parameter tuning algorithms are derived for the CMAC NN controller and control action is employed to achieve a stringent tracking performance. Simulation results are included to justify the theoretical claims even when the network weight elements are initialized at zero.