Welding Path Recognition By Visual And Trajectory Planning Of Robot

"Made in China 2025" clearly pointed out that the future of China’s manufacturing industry to the direction of intelligent and automation. Welding plays an important role in manufacturing, therefore, to implement the intelligent and automation of the welding process, become the inevitable trend of welding in the future development. In industrial application at present, most of the welding robot is conducted by way of teaching and reappearing. One drawback of this way is when external environment changed we need to do again, a low level of welding automation and production efficiency when the production is variety. Welding path by visual identification and trajectory planning technology, a visual sensor is used to identify the welding path instead of the human eye, and the trajectory planning of the welding process, make robot can independently for the purpose of teaching, so as to improve the automation level of welding robot greatly.Based on the high-resolution CMOS camera and UR3 robot, we form the experimental platform for weld image acquisition,as well as the Eye-to-hand relationship. Firstly set up the modeling of visual identify system, and introduces the model and calibration method of camera, using planar checkerboard for camera calibration experiment was carried out, and put forward the camera calibration precision evaluation method based on the external parameters, in order to evaluate the accuracy of camera calibration results. At the same time introduce the robot hand-eye relationship and carries on the solution, get the camera coordinate to robot coordinate transformation matrix.Secondly introduce the welding seam image processing, according to the characteristics of the welding choose Canny edge detection algorithm of welding path for edge detection, and use the Hough transform method to extraction of welding path and get the welding path equation. Then the image coordinates of the characteristic points of the welding path are obtained, which lays the foundation for the following visual control.Construct the torch rectifying control model, and Proportional control law is used as the control algorithm. Through the TCP/IP communication method to realize the communication between computer and robot controller, using cubic spline interpolation method for curve welding trajectory planning, achieve smooth excessive welding process.Finally through the experiment, the welding path recognition accuracy is evaluated, the reason of error in recognition process is pointed out. Through the revised, the absolute error of welding path of x, y, z directions within 0.8 mm, validate the welding path of visual identification and the robot trajectory planning algorithm in this paper is effectiveness.