Design of disturbance reduction control for periodic motion based on error analysis using sensitivity

The arc welding robot uses a weaving motion that periodically oscillates the welding torch installed at the tip of the robot. The weaving motion is used for the arc sensor to follow the weld fusion line, and the error of the oscillation has large effects on the finished shape of the weld metal, so the improvement of motion accuracy is most important foci in arc welding robot control. In this paper, we propose a design method of feedforward control that cancels out the error caused by the effect of Coulomb friction in the periodic motion of the robot and verify its effectiveness using an experimental robot. Based on error analysis using sensitivity to the periodicity of motion, a smooth feedforward compensating torque is derived to reduce the vertical error and the amplitude error, which affects weld quality. In the derivation, the Coulomb friction is regarded as a continuous value and repeated calculations are carried out to calculate feedforward compensating torques that are effective in reducing the effect of friction even when the Coulomb friction coefficients are large. Using an experimental three-link planar manipulator, the effectiveness of the proposed method is verified.