Steel Fusion: Advancing augmented fabrication and robotic sculpting in architectural design

Soomeen Hahm, Leyuan Jiang, Yang Song, Hanjun Kim

Research output: Chapter in Book or Report/Conference proceedingConference Proceedingpeer-review

Abstract

In contemporary architectural practice, the convergence of artificial intelligence (AI), robotics, and augmented reality (AR) is revolutionising how digital and physical processes interact in design and construction. The SteelFusion Pavilion exemplifies a flexible and
adaptive approach to fabrication, leveraging AI-driven optimization, robotics, and AR systems throughout the building process to balance precision, creativity, and material efficiency. By reimagining architecture as an "information system," the project integrates data
generation, processing, and realisation to inform its design-toconstruction workflow. AI algorithms optimised structural geometry and material usage, while robotic processes enabled precise foam block cutting, and AR-guided holographic systems facilitated steel pipe
bending and assembly, demonstrating the synergistic potential of these advanced technologies. Developed in two iterations, Version 1 was a 10-day experimental workshop with students, highlighting the educational value of collaborative learning and exposing challenges such as structural weaknesses and assembly inefficiencies. These
insights directly informed Version 2, a refined prototype constructed by contractors, featuring simplified structural systems, innovative joints, and enhanced constructability. The SteelFusion Pavilion illustrates the transformative potential of dynamically integrating AI, AR, and robotics, offering a model for precision, adaptability, and sustainable innovation in architectural fabrication.
Original languageEnglish
Title of host publicationCAADRIA 2025
Publication statusPublished - 30 Mar 2025

Fingerprint

Dive into the research topics of 'Steel Fusion: Advancing augmented fabrication and robotic sculpting in architectural design'. Together they form a unique fingerprint.

Cite this