Vision-Based Path Generation Method for a Robot-Based Arc Welding System

Abstract

In this paper a vision-based integrated method intended for path generation for a robot-based arc welding system, is presented. The described system is composed of the recently developed pseudo stereovision system (PSVS) or an ordinary stereovision system and the related software. A desired path can be generated, using a part or the entire edge of an image captured from a scene of the robotic environment, a line manually designed in the image, a combination of lines of the previous cases or lines belonging in successive images captured from different scenes. A user can initially process images selecting by means of pull down menus a variety of filters, edge detection methods and operations. Then the desired path as a combination of lines is selected from images. Applying our correspondence algorithm, corresponding edges can be found. Finally, a number of successive path points are calculated by means of the proposed path point calculation algorithm. In on line operation, the vision system mounted on the end-effector can capture images with the desired best view (welding view) of a scene by moving or rotating (using push buttons) the end effector of the robotic manipulator --- PUMA 761. Other facilities of the described system are the selection of a variety of colors and shapes, histogram view, desired magnification, system information and automatic execution of user-selected operations. The graphical user interface is developed in Visual C++, it runs in a personal computer and communicates with the robotic manipulator (PUMA 761) through ALTER communication port.