Invitation to Augmented Carpentry public thesis defense

© 2025 EPFL

© 2025 EPFL

Andrea Settimi will present Augmented Carpentry on Friday, July 4th, 2025 at 16h30 pm (CEST) at EPFL (Auditoire CO2).


The thesis was conducted at the laboratory for timber constructions under the supervision of Prof. Yves Weinand and Dr. Julien Gamerro and is titled: «Augmented Carpentry: Integration of Manual Fabrication into the Digital Value Chain »

Abstract

Timber’s ecological advantages depend on both material selection and production adaptability. While digital fabrication can scale these benefits, centralized, high-tech systems remain costly and exclusive, limiting broader participation among smaller firms and local communities. In Switzerland alone, small and medium enterprises in construction account for over 99 % of businesses and two-thirds of related jobs, underscoring the need for more flexible and accessible digital approaches.

Current efforts to expand timber digital fabrication focus on making high-end robotic systems more user-friendly and adaptive, often through human-robot collaboration. Yet specialized infrastructure stays expensive and complex, restricting widespread adoption. Retrofitting existing assets and workflows has emerged as a more resilient, cost-effective strategy. By upgrading established manual power tools, companies can gain advanced digital capabilities without investing in disruptive systems. This preserves proven hardware, broadens participation, and remains sustainable and scalable. Recent innovations in computer sensing and tracking confirm the feasibility of this retrofitting approach, though technical and conceptual obstacles persist.

This thesis presents Augmented Carpentry, an open-source, free software CV framework designed to integrate manual manufacturing into the digital value chain. Rather than using 2D execution drawings, manual markings, or physical jigs, Augmented Carpentry provides a digital visual assistance system compatible with common cutting and drilling woodworking tools. It leverages affordable sensors, a dedicated XR engine, custom CV components, and a familiar interface to enable real-time tool guidance and direct 3D model visualization in woodworking. By removing conventional analog tasks, this framework seamlessly integrates traditional craftshumanship into a three-dimensional digital workflow.

This framework is initially explored in a pilot study involving a simple drilling task to identify bottlenecks, interface limitations, and potential remedies. While the first generation of Augmented Carpentry proves feasible and scalable, its accuracy requires improvement.

A required development is a custom module for millimeter-precise camera localization, ensuring consistent 3D overlays during fabrication. Its performance is validated against groundtruth data from an outside-in tracking system. Next, a dedicated toolhead localization component employing a simple camera sensor undergoes an experimental campaign to confirm its reliability. Both subsystems are integrated into a cross-platform AR environment built on a UNIX-based XR engine.

To gauge overall system accuracy, we introduce diff-check, a tool that directly compares scanned data with 3D models in a CAD interface. Finally, real-scale timber construction scenarios, using standard circular saws, chainsaws, and handheld drills, demonstrate the complete framework’s millimeter-level tolerances. These results establish Augmented Carpentry as a strong foundation for future exploration of integrating human manufacturing into the digital value chain.

Keywords: timber construction, digital fabrication, cyber-physical fabrication, human-centered technology, extended reality, computer vision.

References

Settimi, A., Gamerro, J., & Weinand, Y. (2025). Augmented Carpentry: Computer Vision-assisted Framework for Manual Fabrication. In Automation in Construction. Elsevier BV. http://arxiv.org/abs/2503.07473

Settimi, A., Hong-bin, Y., Gamerro, J., & Weinand, Y. (2025). ACEngine: Augmented Reality UNIX C++ Engine for Enhanced Visual Guidance in Digital Fabrication. In JOSS: Journal of OpenSource Software. 10.5281/zenodo.14191429

Settimi, A., Chutisilp, N., Aymanns, F., Gamerro, J., & Weinand, Y. (2024). TTool: A Supervised Artificial Intelligence-Assisted Visual Pose Detector for Tool Heads in Augmented Reality Woodworking. MDPI (Vol. 14, Issue 7, p. 3011).10.3390/app14073011

Settimi, A., Yang, H.-B., Gamerro, J., & Weinand, Y. (2024). TSLAM: a tag-based object-centered monocular navigation system for augmented manual woodworking. In Construction Robotics (Vol. 8, Issue 1). Springer Science and Business Media LLC. 10.1007/s41693-024-00118-w

Settimi, A., Gamerro, J., & Weinand, Y. (2022). Augmented-reality-assisted timber drilling with smart retrofitted tools. In Automation in Construction (Vol. 139, p. 104272). Elsevier BV. 10.1016/j.autcon.2022.104272

Settimi, A.†, Gilliard, D.†, Skevaki, E.†, Kladeftira, M., Gamerro, J., Parascho, S., & Weinand, Y. (2025). DiffCheck: a Scan-CAD Evaluation Tool for Digital Manufacturing and Assembly Processes in Timber Construction. ICSA, arXiv. 10.48550/ARXIV.2502.15864



Images to download

© 2025 EPFL
© 2025 EPFL
© 2025 EPFL
© 2025 EPFL
© 2025 EPFL
© 2025 EPFL
© 2025 EPFL
© 2025 EPFL
© 2025 EPFL
© 2025 EPFL
© 2025 EPFL
© 2025 EPFL

Share on