The welding current is approximately propotional to the tip-weldment distance. This property makes the welding torch itself act as a sensor, commonly called an arc sensor. The information from the arc sensor is used to find the groove edge during torch weaving.
Another common automatic welding technique uses vision camera which forsees the groove for which the weld torch is to follow. In this technique, however, noise from the arc light invokes error in identifying the center of the groove and therefore it may deviate the welding path from the actual path.
This research investigate a technique which combines both the arc sensor and vision camera. With such a method the general contour is obtained by the vision camera and the precise seam tracking of the torch is achived through the torch weaving by the arc sensor.
In this thesis, vision system is designed to rotate the camera about the torch axis to endow the tracking ability for an arbitrary path. And the image processing algorithm is proposed so that the vision camera can detect the broad field of view to find the groove center fastly. Also, for the torch weaving, the groove center is found by the detection of the weld current during weaving on the groove. and then, the torch weaves around the groove center found.
The system interrupt is used for using two sensors simultaneously. The subprogram to excute the current sensing and the weaving algorithm is performed by the interrupt request with regular frequency during excuting the image processing algorithm.