Research On The Weld Seam Detection Of A Robot Based On Binocular Vision

Welding is an important technology in the field of manufacturing,and the automatic welding by robot has been widely used in the production process.With the pronouncement of"industry 4.0",the demand for intelligent welding becomes more urgent.At present,the welding robot's trajectory deviation will affect the quality of the weldment due to the lack of feedback characteristics.This paper used binocular vision technology based on the line structured light to get the spatial coordinates of the butt weld,lap weld and fillet weld in the welding path.There were warnings for welding trajectory deviation when contrasted coordinate point sets between the teaching track and detection track.Therefore,it is of great significance to study and design the weld seam detection system.In this paper,the components of the experimental system were calculated and selected,and the spatial positions of the two cameras were mainly introduced.We used the camera resolution and measurement error to get the optical center distance and axis angel of the two cameras.Then we made the vision sensor.After that,the paper introduced camera calibration.We used the Fast operator to extract corner point,and replace this process for the corresponding link of Zhang's calibration tools.It can reduce pixel error effectively.Finally,we can get the results of binocular calibration and hand-eye calibration.Then the pictures were pre-treated,and we used the Hu invariant features of pretreatment pictures as the vectors.We can distinguish three types of weld seam in the line structure light by using BP neural network.In this chapter,we provided cognitive function for different weld calculation.At last,three methods of calculating the pixel coordinates of the weld seam were proposed.On the basis of these,the 3D coordinates of the welding joints were obtained by combining the calibration information and the position of the torch.Then this paper briefly introduced the principles of weld seam detection and deviation warning.And we wrote programs of welding seam detection by the principles.