Structure of Laser Vision Tracking System
The laser, serving as a structured light source, projects laser stripes onto the working surface below the sensor at a predetermined angle. The CCD directly observes the stripes beneath the sensor. The sensor is installed at a predetermined distance in front of the welding torch, allowing it to observe the weld seam. During tracking, welding speed and forward distance are used to calculate the delay time, ensuring that the welding torch progresses along the weld seam.
Laser Vision Sensor
Compact laser vision sensors are typically installed on the welding torch. A laser diode inside the sensor emits a visible laser beam, which, after passing through an aperture, becomes a fan-shaped light band projected onto the workpiece. The sensor contains a CCD camera set at an angle relative to the laser diode, capturing the reflected light from the workpiece surface to image the geometric contour of the joint onto the CCD. Utilizing optical triangulation, horizontal features can be detected, and the height between the sensor and the workpiece can also be determined.
The role of the sensor is to convert the measured physical quantity into a useful electrical signal corresponding to it, to meet the requirements of information processing and control. The control system analyzes and processes the signals from the sensor to obtain control signals. The actuator typically consists of motors and cross sliders, with the servo motor controlled by the control signal.
Image Preprocessing
Image processing in welding robot vision systems can generally be divided into two aspects: first, enhancing, filtering, and binarizing the acquired raw images to remove noise and make the original image clearer with more pronounced edge features, i.e., preprocessing; second, performing edge detection, feature extraction, etc., on the preprocessed images to accurately obtain the center information of the weld seam, facilitating coordinate adjustments by the robot to meet welding quality requirements. A study on real-time image processing methods for visual weld seam tracking has been conducted, initially employing image enhancement to increase image contrast, using median filtering to remove image noise, and extracting the target image from the background image using binarization. In the post-processing phase, key technologies for image processing in laser vision weld seam tracking systems, such as extracting the centerline of micro-light bands and detecting feature points, have been proposed. The centerline extracted using the axial transformation method is single and continuous, and feature points are detected conveniently and reliably using the slope analysis method. This processing method can accurately detect weld seam feature points, with fast processing speed, meeting the real-time requirements of the tracking system.
The above content is an analysis of the structure of the laser vision weld seam tracking system. Before the welding operation begins, the laser vision weld seam tracking system adjusts the position of the welding torch to align it with the center of the weld seam. Then, the power is turned on, the system starts working, and the program begins to run. The CCD in the sensor continuously monitors the image in real-time. A timer is started by the software, adjusting the time interval of the timer according to the welding speed requirements, capturing a frame of images at regular intervals. Through the program, real-time processing of the images is conducted to obtain the center of the weld seam. The deviation of the current weld seam center from the initial weld seam center is used as the detection quantity, and the adjustment amount is output to the robot controller to control the robotic arm to guide the welding torch for automatic tracking.
