Applying image processing for autonomous finding welding line for industrial robot

Abstract

Currently, at the Vietnamese factory, continuous straight-line welding is still done manually or by workers. Some factories have manipulators to do this, but workers have to teach the robot to recognize the weld using a Teach-pendant device. Those jobs will take a lot of time for a welding job as well as uneven product quality. The purpose of the research is to develop an automatic system that combines computer vision and an industrial manipulator with 6 degrees of freedom (6DOF) that can recognize the weld line automatically and command the manipulator to follow the line profile. First, by using a 3D camera and laser system mounted at the end of the 6DOF manipulator and combined with the laser triangulation image processing algorithm, the position of the path from the starting point, sub-point and point weld end are extracted. Then, using an interpolation algorithm, the trajectory is calculated and transmitted to the manipulator. By using this method, the quality of the welded product is guaranteed and also the quantity of welding is increased as the Robot can work from time to time and especially for repetitive tasks. The installation time to weld the product is significantly reduced with the help of the computer vision system. The efficiency of the system is proven through experiments.