Machine Vision for Fiber Placement

Additive manufacturing has gained significant prominence in the production of intricate structural components. Researchers around the world are continuously developing novel processes that utilize polymers and metals to produce three-dimensional objects from computer-aided models. Additive manufacturing processes that combine polymers with fibrous materials have been developed for quite some time; these processes mostly employ short fibers but also increasingly continuous fiber bundles. One such process, wet fiber placement, is currently being developed at Leibniz-Institut für Verbundwerkstoffe (IVW). In this process, a carbon roving is impregnated with a thermoset resin and meticulously guided to the correct position on a moving table by an end effector. The current setup, depicted in Figure 1 (bottom), boasts three translatory and one rotational axes that allow for the manufacturing of 2.5D workpieces. A cutting unit inside the end effector can cut the wet roving to the correct length. The image also shows our current machine vision setup.

A specialized module was engineered to closely monitor the process by using advanced machine vision techniques. The study involved a series of experiments utilizing a webcam and artificial intelligence algorithms, which were integrated into an NVIDIA Jetson Nano platform. This innovative module excels at identifying the initial position of the roving material and accurately measuring both its position and orientation. The data gathered by the module plays a crucial role in regulating wet fiber placement. By analyzing the detected data, the module can help with making well-informed decisions to either initiate or halt the wet fiber placement process as needed.

To facilitate seamless communication and integration with the overall process control system, the module was equipped with various interfaces. These interfaces included an OPC UA server which ensures standardized communication in industrial automation, an optical output for visual feedback, an electrical voltage output for specific control actions and a CSV file output, allowing for detailed data logging and analysis.

In summary, the developed module not only accurately detects the initial position and orientation of the roving material, but also actively contributes to the precise control of the process. Its integration with diverse interfaces enhances its functionality, enabling efficient communication and decision making within the broader industrial workflow.

The project "TopComposite – Topology-optimized and resource-efficient composites for mobility and transport" is funded by the Federal Ministry of Education and Research (funding reference 03XP0259).