2017 Projectile Acoustic Fiber Forest

Page 1

Fibre Acustic Projectile Pavilion University of Tokyo 2017/18



Prof. Yusuke Obuchi (Obuchi Lab)

Professors

Prof. Jun Sato (Sato Lab) Shohei Furuichi

Shuntaro Nozawa

Deborah Lopez Lobato Ao Y ang

Project Management

Hadin Charbel

Mika Kaibara Portugaise

Nathália Barros Abate Rotelli Chen

Shuta Takagi

Second Year Students

Rūta Stankevičiūtė

Xiaoke

Ziyi Wu

Anran Wang Lindstam   Sher Lin Tan Yuqing Shi

Dominika Demlova    Priya Murugeswaran Sofia Iino Cotado

Otto Malcolm

Ruoyu Chen   Wanting Liu

First Year Students

Yuxi Zhu

Aleksandar Kirilov Mladenov

Machiko Asahara

Additional Student Support

Kyosuke Kawamura Taisei Coorporation

Sponsor and Project Collaborator


Statement Can the integration of digital technologies in design and construction processes be used for expressing personal tendencies through the creation of new architectural practices as well as novel aesthetics? The University of Tokyo Digital Fabrication Lab (DFL) has annually been involved in an experimental pavilion installation, aiming at the application of digital technologies to architectural construction. Its projects are the embodiments of human-machine interaction, demonstrating its unique fabrication systems that make a worker behave as part of a 3D printer; translating human understanding of space through sound into design. At the DFL, the role played by digital technologies in design and construction does not necessarily have to make them more efficient and streamlined. Their potential is instead focused on personalising the fabrication process and human natural tendencies. In our view, the act of making as such becomes a design in itself by means of computer assistance. The diversity that humans collectively show is organically expressed in the architecture. PAFF is an architecture made by reference to sound rather than visual information. We developed a sound guidance system whereby a worker’s task is to recognise a sound source in a three-dimensional space, and shoot fibres into various targets. PAFF is expressed through the accumulation of the multiagent system, revealing personal differences in the way that people perceive space through sound.


built pavilion at University of Tokyo campus




positioning on the site


exploded diagram of PAFF


mapping individual sound-space accuracy with virtual sound sources in virtual space

individual profile details

Otto

Domi

Sissi

Ruta


Anran

Carol

Chisato

Domi

Jack

Inoue

Mika

Otto

Pria

Rachel

Ruta

Ryo

Sher

Sissi

Sofi

Natsuki

Yuki

Yuta

individual sound-space maps


Material

Machine Learning Process

The main material in the PAFF pavilion is coconut

The machine learning system consists of three

fibre, also known as coir. It is the natural fibre

parts. First, the data of input angles and actual

extracted from the coconut husk after these husks

shooting angles are imported into the system

are cured to decompose their pulp, then dried (a

to calculate the deviations happened on certain

process called retting) and spun either by hand or

vertical angles.

with the use of machines. Then we used the neural network component Coconut fibre have the highest toughness amongst

incorporated in a grasshopper plugin for computing

natural fibres - even being researched to be used

the correlation between the input vertical angles

as reinforcement in concrete structures. But usually

and the deviations. A curve is generated so we can

it is used as floor mats, brushes, in horticulture and

have a basic understanding of this relationship.

agricultural fields and much more. It is also relatively waterproof and easy to stuck to itself and other

Finally, the shooting range will be remapped from

rough materials thanks to the morphology of the

the real situation to a machine learning corrected

fibre.

range. By applying this modified shooting range, the accuracy of a user is corrected to his/her

Thanks to several tests it was decided that the optimal weight was 4-6g, and that the fibre compressed into a ball was more efficient than when in its fluffy and natural state - although both configurations were used.

projectile of fiber elevation

comfortable range.


STEP01: Point With Deviation

STEP02: Machine Learning Curve

STEP03: Shooting Range Remapping

machine learning to adjust guidance


1

3

2

6 5

4

1 2 3 4 5

-

User Control Panel Vive Tracker Data Input Head Movement Tracking Blowing Tool Movement Tracking Aiming Target Input

7

- Aiming Target Compensation via Machine Learning - Aiming Target Acousticalization

8

- Aiming Output & Deviation Record

6


8 7

grasshopper script


scaled model testing


1:1 testing in lab environment


47 individually handmade meshes


placement of pre-bent columns and shooting of fiber


site with scafold



projectile station


details of fiber canopy







Turn static files into dynamic content formats.

Create a flipbook
Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.