AI-based software from Fraunhofer IGD for assembly line support and quality control

Guided through the assembly process as if by magic

Press release /

Industrial assembly lines vary considerably. Some operate with frequent changes of product setup, while others are characterized by complex processes and the need for intense coordination between frontline workers and backroom experts. Fraunhofer IGD has combined augmented reality (AR) and artificial intelligence (AI) to help users meet the challenges of modern production. The solution consists of an AR-supported assembly workstation that guides employees through the assembly sequence while simultaneously facilitating quality control of components as well as remote support. In this way, manufacturers can reduce error rates and eliminate cost-intensive downtimes. The Fraunhofer team will be presenting the software at the CONTROL trade fair and at HANNOVER MESSE.

Darmstadt: The worker sees the correct sequence and alignment of components through augmented reality glasses – this is not a vision of the future, but already a reality, simplifying processes on existing production lines. The AR-supported assembly workstation developed by the Fraunhofer Institute for Computer Graphics Research IGD makes printed paper instructions superfluous. Instead, four cameras survey the work area and guide human operators through the individual steps via a mobile AR output unit, for example a tablet or AR glasses.

Immediate feedback avoids errors

If the worker picks up the wrong component, a visual error message including an overlay of the correct element appears immediately in their field of vision. Feedback is also provided if the component matches the specification but is positioned incorrectly or the component exhibits a defect such as a missing drill hole. Andreas Franek, deputy head of the Virtual and Augmented Reality department explains: “The software prevents errors in the assembly process from occurring in the first place and from recurring in the future, because if discrepancies are only discovered later, this incurs increased expense and causes cost-intensive downtimes.”

This type of support is already in great demand in the automotive sector and in plant and tool construction as well as in quality assurance. Other possible areas of application include the manufacturing processes for batteries and for individually designed control cabinets. The software offers another advantage, especially for complex product setups: if questions or problems arise, an expert can be summoned from outside without delay to support the person on the assembly line. The expert can explain things in direct virtual presence and project comments in the field of view – and thus on the workpiece itself. Thanks to immersive AR technologies, the expert is able to make a virtual appearance in the work environment, in the user’s experiential space. This allows for natural two-way communication and interaction, as if both participants in the conversation were physically present. This form of telepresence is a core application of the “Industrial Metaverse”.

© Fraunhofer IGD
Image (montage): The AR-supported assembly workstation of Fraunhofer IGD supports production processes and enables remote experts to be connected as telepresence - a core application of the "Industrial Metaverse".

Teaching AI with CAD models

In developing the technology, Franek and his team combined augmented reality with machine learning methods: “The AI is taught with computer-generated images derived from the CAD models. Real photos are not required. This reduces the respective lead time.” In addition, it eliminates the intermediate work step from the CAD models to the carefully structured printed work instructions, which continue to dictate the sequence of assembly in many industrial settings.

Trade fair visitors who are interested in this major innovation can try out the AR-supported assembly workstation for themselves on a machine tool and let the AI guide them through the process. For particularly tricky assembly tasks, they can also receive support via the remote guidance function.