3D control for automatic welding processes of large steel components

Large steel components (> 50 m2) are often welded manually as single pieces. To automate these processes, two tasks arise for measuring technology: the 3D positioning of the welding machine and the process control at the welding seam. For this purpose, Fraunhofer IOF Jena developed a 3D sensor system that can measure both the joining gap and the weld seam with high resolution.


The multispectral 3D sensor technology is attached to a welding robot, which is intended to enable cost-effective and flexible production of steel components, especially for medium-sized companies. Among other things, a mobile cable pull system was developed by the TH Deggendorf, which can replace a cost-intensive stationary portal system.

Mobile welding platform

With this mobile welding platform, the welding robot and the 3D sensor system can be suspended and moved over the workpiece on a multifunctional carrier developed by the project partner Sibau. This results in the following requirements for the 3D sensor system: On the one hand, it must prevent collisions between the platform and the workpiece and, on the other hand, it must measure the welding path exactly. In the course of this, two separate sensor systems were manufactured. While a single sensor is carried on the platform to detect the position and any obstacles on the track, the welding sensor system, which is orders of magnitude more precise, is oriented to the joint gap itself.

The 3D sensor technology

For the orientation of the platform, a time-of-flight (ToF) sensor was chosen, which provides up to 300,000 points/measurements at 15 measurements/s and can measure distances up to 13 m. The 3D data acquired with the ToF sensor can finally be forwarded to a welding specialist, who uses it to define the working area of the welding robot for the actual welding process. Once at the joining gap, the work of the multimodal sensor system developed at Fraunhofer IOF begins. This consists of a miniaturized 3D sensor, which delivers more than 30 3D data records/s with 1 MP resolution each at a measuring field of 400 x 400 mm. Thus, even partly rusty or highly reflective surfaces of the joining partners can be detected and measured in 3D. Depending on the application, the 3D sensor system can be converted for different wavelengths up to the UV and SWIR spectral range.

Robot with sensory measuring systems automatically welds a seam between metal plates.
© Fraunhofer IOF
Multimodal sensor technology on the arm of the welding robot consisting of the welding torch with 3D sensor, a hyperspectral and a thermal imaging camera.

The welding process

Using the 2D and 3D data obtained previously, the welding specialist defines the start and end point of the weld seam in the joint gap. A representation as topography or in false colors facilitates the interpretation of the 3D measuring points. In addition, surface cuts can be determined automatically by the system and suggested as linear welding paths. During the welding process, the process is recorded in 2D with an additional LWIR camera. These and two hyperspectral cameras look at or just behind the weld pool. After the welding process, the seam can be recorded again in 3D with a resolution of 200 μm. Finally, the calibration of all sensors into a coordinate system allows the consolidation and correlation of the multimodal data.


The developed sensor platform was funded by the BMBF as a sub-project "3D sensor systems for small and large scale processes in steel construction - geometry acquisition, path planning and 3D multispectral registration" (FKZ 03ZZ0450) in the joint project "3D multisensor systems and handling systems for automation in steel construction".