The use of laser as a photonic tool in production is industrially established and has led to a change where conventional manufacturing methods are increasingly substituted by laser-based processes. The digitalization of production is pursuing the goal of optimizing productivity as well as the traceability of each individual process step, so that increasingly self-controlled automation and self-regulating processes are becoming possible. The use of cyber-physical systems (CPS) in the field of laser material processing requires a high level of investment, as established solutions are only insufficiently able to meet the high requirements due to the lack of implementation of modern interface concepts, inflexible data models and insufficient data availability, and thus represent cyber-physical islands at best. An evaluation-oriented data management and processing, which focuses on the analysis of laser process data and combines decentralized and centralized concepts, has not been established.
However, such approaches are urgently needed from an industrial point of view, since an enormous number of possible parameter variants exist in laser micromachining, so that a considerable part of the work is required for the identification of process-stable parameters and speed-optimized process solutions. In particular, the evaluation of topographies with features in the nano and micrometer range is often very time-consuming and less automated on the process side, especially when using laser interference-based technology approaches such as Direct Laser Interference Patterning (DLIP) for the functionalization of surfaces. In this area, it has already been successfully demonstrated that AI-based approaches can significantly accelerate process development and help with the technological exploitation of functionalized surfaces (e.g. self-cleaning surfaces modeled after the lotus effect). The data volumes generated during the laser manufacturing process can also be used in particular for predictive analyses, so that sub-optimal quality and machine downtimes can be detected at an early stage of the process. However, this can only be expected in a comprehensive solution with process-integrated data collection, in which the self-learning production system can automatically adapt to the changing context in order to always achieve the optimum in the production process.
The MEDIUS project brings together a multidisciplinary consortium of laser technology, artificial intelligence (AI), human-machine interaction, data communication and surface metrology experts with one vision: Develop a laser manufacturing technology based on Direct Laser Interference supported by Augmented Reality (AR) with an AI-based adaptive expert platform. In this context, Fraunhofer IWS is responsible for the design and development of a DLIP laser processing head, which is paired with an intelligent in-line sensor module. This module will record and analyze the information streams generated during the process, in particular the visual and acoustic emission of the laser ablation. Using suitable control algorithms, active control of the process will then be implemented. In addition, predictive modelling of surface functionalities after laser processing, based on the above and other machine data, will be developed. This will allow laser-induced surface functionalities, such as the defined water contact angle, to be predicted on a material-specific basis, thus reducing the development times for complex surface functionalities. The project addresses the development of a photonic predictive manufacturing system based on several coupled AI-based prediction systems for the surface laser processing of the future.