Self-learning welding robot

Innovation project: Intelligent processing of large parts with high tolerances

Robotic technology is used in many industrial production processes to speed up processing and minimize costs. At present, welding robots must be “taught”; this means that the movement sequences are learned or programmed in advance. These processes require welding machines to be highly accurate in their positioning, which often cannot be guaranteed – particularly when working on large parts with high tolerances, such as steel girders for the construction industry. As a result, the only way to ensure the necessary quality involves additional work – either manual rectification or readjustments to the equipment.

The aim of the project is to develop an intelligent welding robot that can reliably process large parts with high tolerances. Using innovative detection and control technology, the robots should check the individual properties of the part and adjust the position of the robotic arm accordingly. This makes it possible for large parts to be processed automatically with a high degree of precision and removes the need for manual rectification or readjustments to the welding machine.

To that end, optical detection systems such as image processing and pattern recognition are developed to allow the robot to quickly and reliably check the actual proportions of the part and compare them with the target position. New software algorithms are also developed to calculate the optimum target position for the robotic arm and the paths required to reach it. In the process, potential collisions with the workpiece and other robots must be identified and avoided. The software relays the calculated parameters to innovative actuators in the robotic arm, where the movements are carried out in real time according to specifications. The project draws on the results of cross-sectional projects in energy efficiency and intelligent networking. The process is tested using a demonstration system, and then integrated into the manufacturing process.

Intelligent welding robots allow large parts to be processed accurately and reliably. There is no need for manual post-processing or readjustments to the robot, which improves the efficiency and effectiveness of the manufacturing process. It is expected that programming and post-processing work can be reduced by 50%. The results can also be transferred to other manufacturing processes such as the automated application of sealants and adhesives to large parts.

Project duration
01 July 2013 - 20 June 2016