J. Eur. Opt. Society-Rapid Publ. 21, 30( 2025) 319
the two images was 31.54 %, which is very high for a process like polishing with a lot of parameters. This speaks for the high quality of the set up and the meaningfulness and feasibility of the idea.
The interferometer used is a self-made one and actually intended for larger components. Due to the large measuring cavity and a suboptimal installation location, the measurement is in the range of measurement uncertainty. Therefore, the interferometric measurement result in particular should be treated with caution.
The remaining process inaccuracies of the intelligent polishing lever result from ignored process parameters: polishing pressure, vibrations, temperatures, density, pH-value or others.
Nevertheless there are some disadvantages in the setup: Most machines are loaded from above, often with a crane. The camera must be positioned over the machine in a repeatable manner, but in most cases it must be movably attached using a crane, etc. The camera may need to be mounted differently, maybe adjustable.
6.1 Incorporating polishing pressure
While this study focuses on the influenceofrelativevelocity on material removal, polishing pressure remains a critical yet unmonitored factor in the process. The Preston equation clearly indicates that pressure directly contributes to the removal rate. However, in most lever arm polishing setups, the pressure applied to the workpiece is neither constant nor well characterized. It is typically the result of dead load, manual adjustment, or geometric constraints such as contact point location and lever configuration.
To advance toward a fully predictive and automated process model, it will be necessary to implement a method for measuring and, ideally, controlling the applied polishing pressure. This may include the integration of force sensors at the tool-workpiece interface, or the development of indirect estimation methods based on deflection, torque, or surface contact analysis. Future work should investigate how pressure varies during operation and how it interacts with velocity to affect material removal. A combined analysis of these parameters would significantly improve the reliability of removal prediction and enable real-time process adjustment.
7 Conclusion
This publication presents a practical investigation into the correlation between relative velocity and material removal in lever arm polishing. A sensor-based system was developed to track and quantify the kinematic conditions during polishing, enabling a spatially resolved analysis of the polishing process. Rather than proposing a complete process or control strategy, the work focuses on validating whether relative velocity can serve as a meaningful predictor for material removal under real production conditions. The experimental results confirm a measurable correlation, indicating that such an approach could form the basis for future, modelbased process automation. This lays the groundwork for further investigations into additional polishing parameters, such as pressure distribution or contact area, which are essential for accurate removal modelling.
7.1 Overlay
If not only the relative speed is considered, but also the contact surface, new possibilities arise with the structure. The Preston equation can be applied to each individual position( here: image capture) and used much more precisely. Also with regard to the coverage of the work piece with the polishing tool, that enables a better prediction of the material removal. However, this requires precise pressure control.
In the future, the technology shown here can also be used profitably for a known technology. Instead of the relative speed, the coverage of the polishing foil with the work piece can also be used to calculate the material removal. Flower-shaped polishing sheets were originally used for concave and convex polishing tools. In the past, flat work pieces were also processed with it. The uncovered area supports the supply of polishing agent. In the past, trained specialists learned how to design such tools, but the knowledge is gradually being lost. Use of computers allows reuse of such process capabilities. Unfortunately, the use does not reduce one of the many process parameters, but rather adds a new, albeit controllable, parameter.
Mathematically, the calculation for all shapes is quite complicated. By using optical measurement methods, the area can be easily approximated depending on the radius and also suitable for production. Different shapes and the corresponding coverage are visible in Table 2. Ontheleftsidesuchpolishingflowers are shown, on the right the relative surface area depending on the radius. This can be done in a similar way with regard to the area load.
7.2 Machine learning
This paper opens the door to complete automation of polishing lever arm machines. If the pressure can now be kept constant or regulated, the material removal can be calculated precisely and the structure can be 100 % automated. Measuring and regulating the pressure in the polishing process is a challenge and not easy to implement [ 22 ]. 100 % monitoring, ML Model, calculation of the optimal removal and support of operators can be seen as an intermediate step. The machine learning model will also train process errors and due to that calculate a more precise material removal.
The set up can be also useful in the automation of CCP: camera detects on a lever arm a functional process, which is later transfer to a sub aperture CCP process. There are also publications where this set up seemed useful for their ideas and solutions [ 13 ].
For the future: The camera and other sensors created the basis for big data analyzes and machine learning. Generating of data will be the main focus of data driven solutions in the future. For a good machine learning model, it is not the quality of the neural network that is crucial, but rather the number of data sets [ 23, 24 ]. With the current setup, many data sets can be generated in a timely manner