Skip to content
Technical University of Munich

VAULT-AI

Computational design and fabrication of self-supporting VAULTed structures through AI-driven human-robot cooperation

Principal Investigators 

Prof. Dr. Pierluigi D’Acunto,
Professorship of Structural Design, TUM School of Engineering and Design

Prof. Dr. Kathrin Dörfler
Professorship of Digital Fabrication, TUM School of Engineering and Design

Prof. Dr. Stefan Leutenegger
Professorship of Machine Learning for Robotics, TUM School of Computation, Information and Technology 

Prof. Dr. Angela P. Schoellig
Chair of Safety, Performance and Reliability for Learning Systems, TUM School of Computation, Information and Technology 

Motivation and Goals Vaulted structures use distinctive double curvature to bear loads with minimal material. While conventional design focuses on the final structure, assembly often requires extensive temporary supports, especially for large spans. Our project employs novel form-finding and self-supporting methods combined with robotic fabrication, enabling human-robot teams to build with precision and temporary support the structure during assembly, reducing waste by eliminating scaffolding. Integrating 3D perception and spatial AI ensures seamless collaboration between mobile robots and human builders, with AI-based action recognition providing safe robot guidance.

Expected Results The effectiveness of the proposed methods will be validated through experimental case studies featuring two self-supporting masonry vaults. These experiments will illustrate how an integrated AI-driven human-robot system can transform the construction process, significantly enhancing material efficiency, flexibility, and quality control.

Prior Research

D’Acunto P. et al.: “Vector-Based 3D Graphic Statics” International Journal of Solids and Structures 167: 58–70, 2019.
Dörfler K. et al.: “Mobile Robotic Brickwork” Robotic Fabrication in Architecture, Art and Design 2016,204–2017, 2016.
Schaefer S. et al.: “GloPro: Globally-Consistent Uncertainty-Aware 3D Human Pose Estimation& Tracking in the Wild”, arXiv, 2023.
Heins A.. et al.: "Mobile Manipulation in Unknown Environments with Differential Inverse KinematicsControl." 18th Conference on Robots and Vision (CRV), 64-71, 2021.
Proposed computational design and fabrication of self-supporting vaulted structures leverage AI-driven human-robot cooperation. Human builders and mobile robots collaborate in the same workspace, providing an alternative to fully automated construction. (Credits: Frederic Chovghi, Professorship of Structural Design)
To top