REFUGE 2.0 - Artificial Swissness

LDM

In this studio, we will question the role of artificial intelligence and deep learning in architecture. Is creativity in architecture only exclusive to humans? Can machines automatically learn and generate meaningful architecture? Can they go beyond quantifiable data and optimization and enter the realm of architectural quality? Can machines capture and generate something as fragile and elusive as “Swissness” in architecture? What are the dangers and dark sides when we let the machines loose?

The object of the studio is the alpine refuge, an architectural typology located in the high mountains, accessible only by foot/ski or by air, ordinarily to provide shelter, food and accommodation. The Swiss Alpine Club (SAC) currently operates 153 alpine refuges around various peaks and mountain paths. We will use machines to capture, collect, and learn the architectural DNA of the vernacular characteristic, in our case the “Swissness” of alpine refuges – and critically engage AI to design a new alpine refuge that embodies the quality of that very Swissness.

The first part of the semester will consist of curating data and manipulating a GAN (or equivalent deep learning network) to create variations of artificially generated alpine refuges with the quality of “Swissness”. The collection of facades of Swiss alpine architecture will serve as samples for the algorithms.

During the second part of the semester, students will select and interpret the output generated by the algorithm, and, through a process of inverse geometric projection, transform the machine-generated images into 3D architectural forms.

The final part will assess and refine the results in relation with the extreme climatic factors prevalent in the alpine context. Students will be asked to define the size, location and precise program during the semester, and may consider the use of drones for various purposes including the delivery of supplies.

The studio will culminate in a personal interpretation of the Alpine Refuge as a final project: a privileged, self-sufficient space of retreat and shelter, a moment of peace and serenity, amidst a changing world and confused ecology.

Files
Description Atelier Huang Spring 2020 (PDF)
LINKS
Team
Unit:
LDM
Teachers:
Jeffrey Huang
Assistants:
Christina Doumpioti, Georg-Christoph Holz, Mikhael Johanes, Frederick Chando Kim
Infos
Year:
2020
Period:
Master, Spring
Category:
Course
Topic:  
Architecture, Construction, Experimentation, Heritage, History, Technology
Copyright:
CC BY Licence
Permalink
livingarchives.epfl.ch/projects/4940/refuge-20-artificial-swissness/