This year's research question is âHow can we support archaeologists and researchers in your field?â Each team must now identify a problem related to this research question and develop an idea or solution for this problem.
The problem we have chosen to tackle this year is the manual search for matching shards. This takes archaeologists an enormous amount of time. Our goal is to optimize this process using technology. We are trying to solve this by developing a low-cost 3D scanner that digitizes finds. Local AI then virtually assembles the pieces and creates precise three-dimensional instructions for researchers to reconstruct them.
The 3D scanner
Our scanner uses photogrammetry to combine maximum precision with economic efficiency. Based on the OpenScan model, we have developed an optimized solution that is a cost-effective alternative to expensive industrial devices.
The device captures objects up to 20 x 20 cm in an almost complete 360° radius. The multitude of overlapping individual images creates a detailed point cloud that digitally maps the object with an accuracy of up to 0.1 mm (at 1440p and 350 images). To perfect the process for archaeology, we have also modified the software and interface. This makes the system not only intuitive to use, but also optimally prepared for subsequent AI processing.
Original model by OpenScan
We were inspired by this model, as our approach is based on a similar principle. The accuracy is ~0.01 mm, and objects with dimensions of up to 9 x 9 x 9 cm can be scanned.
Version 1 (partially OpenScan)
This model is an earlier prototype of the OpenScan project, which we experimented with and adapted to our requirements. This version also supports larger cameras, such as DSLRs or mirrorless system cameras. It can also be used to capture larger objects, although the accuracy decreases slightly.
Version 2 (first self made version)
Since we reached technical limits with the previous version (v1), we decided to design our own model. Instead of moving the frame, we now use a slide on which the camera is mounted. The advantages are lower moving mass and greater stability. The system is also scalable: in our current version, it is possible to scan objects and fragments up to a size of 20 x 20 x 20 cm. We achieve an accuracy of ~0.1 mm.
Version 2.5
Subversions of our 2nd version
Version 2.1
Basic construction with slide
Version 2.2
Enlargement of the slide's gear ratio
Version 2.3
First use of lighting (produced in-house at the school)
Version 2.4
Single-color background for better 3D scan quality, minor improvements to the electronics
Version 2.5
Change of background color for better contrast
Version 2.1
Version 2.3
Version 2.2
Version 2.4
Version 3
To increase user-friendliness and enable the 3D scanner to be operated without an internet connection, we have added a control panel. Since the illumination in the second version was still uneven, the lighting concept has been optimized. The artifact is now illuminated not only from the front, but also from all four corners. The carriage has also been fundamentally redesigned: The motor is no longer located in the carriage itself, but together with the electronics in the base. This saves weight and allows for the flexible use of different motor types. In addition, the 3D scanner has become modular. When disassembled, it fits into a 100x40x40cm box, including the wooden base (if you squeeze it in properly, you can even fit some tools in there đ).
We have also switched the electronics from the original OpenScan âGreenShieldâ to our own development. This offers us the following advantages: an adapted base voltage, connections for 24V LEDs, three instead of two motor drivers (as the carriage has two drive motors), PWM dimming of the LEDs, and direct access to the remaining GPIO pins.
Version 4
The v4 is still in the works at the moment :)
Starting with the third version, we developed the motherboard(s) ourselves. This allowed us to tailor them to our needs and gave us flexibility in development. Mr. Nussbaumer kindly allowed us to manufacture the spotlights for the lighting ourselves (using our school's internal etching equipment) in the electronics workshop.
etching machine
The AI
---
The added value in archaeology and museologyÂ
The possible applications of our 3D scanner are particularly diverse and forward-looking in the fields of archaeology and museology. One key advantage lies in the support it provides to archaeologists: the precise digital images make the laborious manual task of piecing together broken shards considerably easier, as adjustments can first be tested virtually and non-destructively.
Museums also benefit from digital recording, as they can now create high-quality duplicates of their exhibits at low cost. These copies make it possible to keep valuable originals safe while making the replicas accessible to the public in exhibitions. This digital approach also allows archaeological artifacts to be shared across museums, thereby promoting international research and collaboration.
A key economic advantage is the use of open-source technology, which means that institutions do not incur any software or license fees. Since the system also functions independently and without an internet connection (offline operation), it offers maximum discretion and data security. This is particularly valuable for ensuring that documentation can be carried out smoothly and securely, even in the case of secret or sensitive excavations.
According to Annette Schiek, AI offers a solution to a core problem in archaeology: the virtual reassembly of fragments that are physically scattered across different museums around the world. She also believes that this digital method is scalable and can be applied beyond ceramics to large-format artifacts such as statues or buildings.
Special thanks go to Mr. MĂźller, who approached our project with great enthusiasm from the very beginning. He provided us with invaluable support in integrating the archaeological perspective into our research. Thanks to his in-depth expertise, we were able to better understand the requirements of modern archaeology and tailor our scanner specifically to the needs of fieldwork and artifact evaluation.
Mr. Ăzçelik's support was crucial both in presenting our 3D scanner and in developing potential application scenarios in museums and at archaeological sites.
Thomas Megel's in-depth expertise was invaluable both in the conception of the general scanner design and in the implementation of our specific optimizations. In addition, he made a significant contribution to benchmarking, where his support and expertise were crucial to the success of our tests.
Mr. Kubitschke provided us with significant support in creating the circuit diagrams and in the overall development of our in-house electronics. In addition, he made a decisive contribution to the implementation of a functioning polarization of the lighting units.
Special thanks go to Professor Chen Feng, who supported our project from the outset with great praise and encouragement. He and his entire team were enthusiastic about our initiative to adopt their technological approach and develop it further in an innovative way. They expressly welcomed this form of scientific exchange. The team, and above all Mr. Jiang Zeyu, provided us with prompt and competent support at all times, particularly when it came to complex issues relating to the implementation of artificial intelligence and the optimization of the program code.
Acknowledgment
This project was an intense journey that we could not have accomplished without the valuable support of our coaches and numerous experts. Without their tireless efforts, this result would not have been possible in this form.
We would like to express our sincere thanks to our specialist supervisors: Professor Chen Feng and his team (especially Jiang Zeyu) for their support with AI implementation, and Mr. MĂźller, who helped us to integrate the archaeological perspective and field requirements deeply into our research. We would like to thank Mr. Kubitschke for his expertise in electronics development and light polarization, while Mr. Ăzçelik made a significant contribution to precisely identifying the use cases for museums and excavations.
We would particularly like to highlight Thomas Megel, whose in-depth expertise in the conception of the scanner design and critical benchmarking was invaluable to the success of our tests. We would also like to thank Annette Schiek (Textile Museum Krefeld) for her valuable input; her assessment that AI enables the virtual merging of scattered fragments has significantly reinforced our vision of scalable, digital archaeology.
Finally, we would like to thank all those involved who have made a decisive contribution to this project through their expertise and trust in our work.