Development of Groove Recognition Algorithm with Visual Sensor

Practical Development of Visual Sensor for Welding Robot (Rep. 1)

Abstract

It is necessary to develop excellent sensors in order to improve the automation or the artificial intelligence of the welding operation.
Laser visual sensors are designed to recognize the configuration or debect the position of a weld by irradiating a slit light on to the weld joint, receiving the light with a CCD camera, and then converting the two dimensional information into three dimensional information by using image processing.
The visual sensors for the welding robot must be free from the influence of welding process disturbances, such as light, heat, spatter and fumes.
This paper deals with optical component arrangements and image processing algorithms. The optical component of a sensor head helps to avoid optical window contamination caused by spatters or fumes. The image processor can process image data very quickly and accurately, even if the welding arc light illuminates the surface of the workpiece. The accuracy of detecting the position is within ±0.2 mm.