Pilot line B: Optics

High reject rate in the production of optical systems

Optical components, such as camera or projection modules, consist of several high-precision individual components (lenses, beam sources, sensors, etc.). Only the correct combination of all individual components ensures functionality, for example in the form of a sharp image. If only one component is defective, the entire system will fail to function. Due to unstable production processes, a significant proportion of the optical systems produced are rejects.

Digital twin along the entire production chain

The aim of the optics pilot chain is to optimize the manufacturing process of optical components on the basis of data and to significantly reduce rejects. To this end, measurement and simulation data will be collected during the course of the project from all links in the production chain (optics design, production by means of glass forming and optics assembly), which will be combined to form a comprehensive digital twin.

An important goal of the project partners is the uniform data flow, which should improve the exchange between the individual chain links in the pilot chain. A data lake platform developed in the project serves as the data pool. All the data relevant to the digital twins is available there, and users can store their own data.

Less scrap through digital linking of optical design and glass forming

In the "Optics Design" section of the project, approaches are being developed to ensure a high yield of usable optics systems. To this end, the lens arrays created in the optics design are first manufactured virtually using glass forming. The Finite Element Method (FEM) is used for the forming simulations. The data collected in this process is prepared and processed into a digital twin. Since the data is stored on the Data Lake platform, it is available to subsequent optics designs and optics assembly and is incorporated into the processes.

Facilitated variation in optics assembly

The goal of the "Optics Assembly" project section is to develop and investigate self-learning, adaptive adjustment processes for automated optics assembly, following the example of nature. The self-learning processes developed in the EVOLOPRO project are based on neural networks as well as the use of algorithms and are trained and adapted by means of the data generated in the optics design as well as during manufacturing on the machine. The digital results are made tangible through the production of demonstrator components.