Machine operators still require extensive knowledge and experience in order to produce highly sophisticated components. At the same time, processes are becoming more demanding as a result of material savings in lightweight construction, for example: In the case of thin-walled components, component vibrations during machining are more likely. This, in turn, can cause anomalies in the workpiece and lead to component rejection. Since operators have had only few or no assistances, they often had to rely on their sensory perception to monitor and control such manufacturing processes. They usually control the stability of the process through their sense of touching and hearing; when they interpret the sounds of machine vibrations or assess the surface quality of the finished parts by looking at the component in detail, for example.
If process data is recorded in real time and automatically evaluated, future scenarios can be derived using analytical and data-driven models. If this data is provided to the operator in the form of intuitively interpretable information, the production can be significantly simplified. Intelligent assistance systems, such as those widely used in the automotive industry, offer an easy way towards stable and efficient manufacturing processes. They can increase the usability of the entire system and enable that even an inexperienced operator can operate the machine in optimal way.
The aim of the "Glassist" project is to combine machine learning algorithms and manufacturing technologies for an intuitive visualization of the process condition. For this, the project partners are developing an assistance system for the machine operator that uses augmented reality. The system provides process-accompanying information such as recommendations for action and details of chatter marks regarding the current tool position to support the machine operator. This increases the user-friendliness of the machine and its process transparency, thereby reducing the time and expense involved in subsequent quality controls.
In addition to project coordination, the Fraunhofer IPT implements various demonstration scenarios and verify them on the applications defined in the project: Detecting deviations in machine data, tool-side chatter, tool wear, imbalances and collisions, process-dependent correction of control parameters and monitoring of the linear scales. For this purpose, sensors are integrated in two machines tools and networked with a database. The researchers are focusing on the quality of the recorded reference data, the temporal synchronicity, the analysis and the labeling of the data, low measurement effort and an intuitive display (HMI) of quality-critical information for machine operators.