At the second "AI for Laser Technology" conference on September 28 and 29, 2021, the current status of the application of artificial intelligence in laser material processing will be discussed.

Researchers, plant engineers, software developers and mechanical engineers will meet here to exchange ideas. In addition to the specialist presentations, the laboratories at the Fraunhofer Institute for Laser Technology ILT will be open for virtual tours. Due to the current regulations, the conference will take place online.

At the first "AI for Laser Technology" conference two years ago, the question was still where artificial intelligence (AI) could be used in manufacturing. In the meantime, development has progressed rapidly and many ideas have arrived on the shop floor.

Various trends are coming together: on the user side, for example, there is a desire for 100% quality control. While weld seams in automotive production were previously only checked with random samples, AI now makes it possible to monitor the quality of each individual part in-situ with high precision and quickly identify potential production faults. This requires in-line process diagnostics, which have been developed in recent years. It is now the task of AI to process the resulting data volumes in real time, which has only become possible with modern computing technology. "At this year's conference, we want to look at and discuss precisely these trends and the state of the art," says Peter Abels from Fraunhofer ILT in Aachen, describing the aim of the conference organized by his team.

Collecting complex data and using it for control

There are various options for using AI in production: This often begins with the evaluation of image or other data. With a human "teacher", AI can also recognize structures in complex data. This allows deviations from a predefined optimum to be detected at an early stage. Process control becomes possible. The highest level is reached when the recording and processing of data is combined with control to form an autonomous process.

Co-organizer Christian Knaak from Fraunhofer ILT will demonstrate this using a specific example: "We were able to significantly improve the results of metallic 3D printing using AI." To this end, the surface of each layer of the component was photographed using a high-resolution HDR camera in a system for the Laser Powder Bed Fusion (LPBF) process. Two effects can be captured with the image data: On the one hand, any distortion of the component during the process can be measured. On the other hand, the roughness on the surface can be checked. In this way, defects can be classified during production. With the help of AI, laser parameters can also be changed during the process in order to react dynamically to changes in the process status. The quality of the parts is improved and defects are prevented before they occur.

The presentations at this year's "AI for Laser Technology" conference will therefore focus on the acquisition and processing of data from production processes such as laser welding. A key objective here is end-to-end process and quality control.

A second focus is on AI-based control processes. The development of AI software in various applications will also be examined. AI not only enables production processes to be optimized in the direction of zero-defect production. In processes with a lot of complex data, such as the development of modern optics, it also provides a reduction in complexity. This makes development processes clearer, more determined and less dependent on the intuition of individual experts.