Welding, as an important connection method in modern manufacturing, has increasingly penetrated into the processing of various types and varieties of materials in various industries and fields. In the traditional manufacturing field of the past, handheld welding methods were mostly used, which had relatively high requirements for welding workers and low welding efficiency, making it difficult to meet the needs of modern welding production. With the emergence of semi-automatic welding technology, the combination of machines and manual labor has improved the degree of welding automation to a certain extent, greatly improving the welding production environment and also increasing the welding production efficiency to a certain extent.

With the accelerated transformation of manufacturing towards digitization and intelligence, the measurement and positioning technology of welds is also constantly being updated and upgraded. In this issue, Xiao Ming will share the use of Akusense visual sensors in the presence/positioning of welds.
Application scenario
The tested object is a circular metal workpiece
▪ Check for welds on circular metal
▪ After detecting the weld seam, guide the positioning robotic arm to grab it

Selection analysis:
1. There is a large amount of welding slag on the metal surface, which can easily interfere with the sensor and cause misjudgment
2. The sensor is installed on the robotic arm and rotates towards the circular ring for motion recognition. The sensor is required to be able to move for recognition and has a certain speed requirement
3. The shape of the weld seam is not uniform, and the weld seam of each product is different. It should be compatible for use
4. Customers want to save costs, reduce costs, and increase efficiency, so they need to choose affordable solutions
solution

Visual Sensor VDS20 Series
1. Select Meiji Vision standard product VDS20 series, equipped with deep learning algorithm, which can ignore the influence of welding slag
2. The frame rate of VDS20 is 60FPS, making it easy to handle mobile recognition scenarios
3. Equipped with high-resolution image sensors and high-speed image processing technology, it can capture more details, improve detection speed and accuracy
4. By uploading successful and failed samples of deep learning, sensors can detect more accurately
imaging effect

Detected weld seam, successful positioning OK

No weld seam NG


NG has no interference from welding seam water pouring, and the detection effect is normal
