XRDS

Crossroads The ACM Magazine for Students SPRING 2016 V OL .2 2 • NO. 3 XRDS.ACM.ORG

Digital
Fabrication
A Manifest for
Digital Fabrication
Soft Printing with Fabric
Fabrication Lends a Hand

Are you looking for
your next IT job?
Do you need Career Advice?
The ACM Career & Job Center offers ACM members a host of
career-enhancing benefits:
• A highly targeted focus on job
opportunities in the computing industry
• Access to hundreds of industry job postings
• Resume posting keeping you connected
to the employment market while letting you
maintain full control over your confidential
information
ACM’s Career
& Job Center
• Job Alert system that notifies you of
new opportunities matching your criteria
• Career coaching and guidance available
from trained experts dedicated to your
success
• Free access to a content library of the best
career articles compiled from hundreds of
sources, and much more!

Visit ACM’s Career & Job Center at:

http://jobs.acm.org
The ACM Career & Job Center is the perfect place to
begin searching for your next employment opportunity!

Visit today at http://jobs.acm.org

 World-Renowned Journals from ACM
ACM publishes over 50 magazines and journals that cover an array of established as well as emerging areas of the computing field.
IT professionals worldwide depend on ACM's publications to keep them abreast of the latest technological developments and industry
news in a timely, comprehensive manner of the highest quality and integrity. For a complete listing of ACM's leading magazines & journals,
including our renowned Transaction Series, please visit the ACM publications homepage: www.acm.org/pubs.

ACM Transactions ACM Transactions

on Interactive on Computation
Intelligent Systems PLEASE CONTACT ACM MEMBER
Theory SERVICES TO PLACE AN ORDER
Phone: 1.800.342.6626 (U.S. and Canada)
+1.212.626.0500 (Global)
Fax: +1.212.944.1318
(Hours: 8:30am–4:30pm, Eastern Time)
Email: acmhelp@acm.org
Mail: ACM Member Services
General Post Office
PO Box 30777
New York, NY 10087-0777 USA

ACM Transactions on Interactive ACM Transactions on Computation

Intelligent Systems (TIIS). This
quarterly journal publishes papers
on research encompassing the
design, realization, or evaluation of
interactive systems incorporating
some form of machine intelligence.
Theory (ToCT). This quarterly peer-
reviewed journal has an emphasis
on computational complexity, foun-
dations of cryptography and other
computation-based topics in theo-
retical computer science. www.acm.org/pubs

PUBS_halfpage_Ad.indd 1 6/7/12 11:38 AM

 Crossroads
The ACM Magazine for Students
SP RING 2 01 6 V OL . 2 2 • NO . 3

14

begin
5 LETTER FROM THE EDITORS

8 INBOX

9 INIT
Digital Fabrication:
A human-machine interface for
advanced manufacturing
By Stefanie Mueller
and Nadya Peek

11 ADVICE
How to Deal with Negative Critique
By Andy J. Hunsucker

12 UPDATES
RIT SIGCHI
Cover by Andrij Borys Associates,
By David Byrd
using photo by Steve Marsel,
courtesy of Nervous System
14 CAREERS
How to Get Hired at a Startup
as a Software Developer
By Jack Minardi

15 MILESTONES
Additive Manufacturing
Top Image by Dotshock / shutterstock.com

By Jay Patel

16 BLOGS
Exploring Virtual Reality—
Are We There Yet?
By Andrew J. Hunsucker

2 XRDS • SPRING 2016 • VOL.22 • NO.3

 Crossroads
The ACM Magazine for Students
SP RING 2 01 6 V OL . 2 2 • NO . 3

DIGITAL FABRICATION

38 64 86

features end
22 FEATURE 54 FEATURE 78 LABZ
A Manifest for Lots of Parts, Lots of Formats, Design Informatics Lab
Digital Imperfection Lots of Headache By Feras Alsaggaf, Javier Villarroel,
By Amit Zoran By James Coleman, Craig Long, and Billy Wong
Andrew Manto, and Trygve Wastvedt
28 FEATURE 79 BACK
Personal Fabrication: 58 FEATURE The Emergence of 3-D Printing
From automated machines Drowning in Triangle Soup: By Asmaa Rabie
to augmented tools The quest for a better 3-D
By Ilan Moyer printing file format 81 HELLO WORLD
By Jesse Louis-Rosenberg The Brownian
32 FEATURE Wanderlust of Things
L-to-R: Image by Andrey Nikolaev; Painting by Rick Guidice courtesy of NASA; Image by Leonid Andronov

Creating Animated Characters 64 FEATURE By Marinka Zitnik

for the Physical World Island Three Revisited: O’Neill
By Stelian Coros cylinders and digital materials 84 ACRONYMS
By Daniel Cellucci and
38 FEATURE Kenneth C. Cheung 84 POINTERS
Building a Toolkit for
Fabricating Interactive Objects 68 FEATURE 86 EVENTS
By Andrew Spielberg, 3-D Printing: Green or not?
Alanson Sample, Scott E. Hudson, By David Rejeski 88 BEMUSEMENT
Jennifer Mankoff, and James McCann
70 FEATURE
44 FEATURE Fabrication Lends a Hand:
3-D Printing Interactive Objects Creating custom
By Valkyrie Savage assistive technology
By Erin Buehler
50 FEATURE
Soft Printing with Fabric 76 PROFILE
By Huaishu Peng, Scott Hudson, Dennis Bormann: The man
Jennifer Mankoff, and James McCann who introduced Antarctica’s
Davis Station to 3-D printing
By Adrian Scoică

XRDS • SPRING 2016 • VOL.22 • NO.3 3

COMMUNICATIONSAPPS
E DI T ORI A L B O A RD
Editors-in-Chief
Jennifer Jacobs
MIT, USA
Okke Schrijvers
Stanford University, USA
Departments Chief
Adrian Scoică
University of Cambridge,
UK
Issue Editors
Stefanie Mueller
Hasso Plattner Institut,
Germany
Nadya Peek
MIT, USA
Issue Feature Editor
Numair Khan
New York University, USA
Access the Feature Editors
Judeth Oden Choi
Carnegie Mellon
latest issue, University, USA
Richard Gomer
University of
past issues, Southampton, UK
Shudong Hao
BLOG@CACM,
University of Colorado
Boulder, USA
Talia Kohen
News, and Bar Ilan University, Israel
Nidhi Rastogi
Rensselaer Polytechnic
more. Institute (RPI), USA
Billy Rathje
University of Oxford, UK
Yang Shen
UCLA, USA
Department Editors
Abhishek Bhattacharya
Amity University, India
David Byrd
University of Baltimore,
USA
Nur Al-huda Hamdan
Available for iPad, RWTH Aachen University,
Germany
iPhone, and Android
Andrew J. Hunsucker
Indiana University, USA
Tejas Khot
Mumbai University, India
Bryan Knowles
Western Kentucky
University, USA
Finn Kuusisto
University of Wisconsin-
Madison, USA
Darshit Patel
Pimpri Chinchwad College
Available for iOS, of Engineering, India
Jay Patel
Android, and Windows University of California
Berkeley, USA
Asmaa Rabie
Cairo University, Egypt
Johanna Schacht
Karlsruhe Institute of
Technology (KIT), Germany
Marinka Zitnik
University of Ljubljana,
Slovenia
4 XRDS • SPRING 2016 • VOL.22 • NO.3
Digital Content Editor A D V IS OR Y B O A RD SUB S C RIBE
Pedro Lopes
Hasso Plattner Institut, Mark Allman, Subscriptions ($19
Germany International Computer per year includes XRDS
Science Institute electronic subscription)
are available
Web Editor Bernard Chazelle, by becoming an
Abhineet Saxena Princeton University ACM Student Member
Guru Gobind Singh www.acm.org/
Indraprastha University, Laurie Faith Cranor,
Carnegie Mellon membership/student
India
Alan Dix, Non-member
Lancaster University subscriptions:
News Editors
Marius-Tudor Benea $80 per year
David Harel, http://store.acm.org/
UPB, Romania and UPMC, Weizmann Institute
France acmstore
of Science
ACM Member Services
Doris Lee Panagiotis Takis Metaxas ,
University of California, To renew your ACM
Wellesley College membership or XRDS
Berkeley, USA
Noam Nisan, Hebrew subscription, please send
University Jerusalem a letter with your name,
Bloggers address, member number
David Byrd Bill Stevenson , and payment to:
University of Baltimore, Apple, Inc.
USA ACM General Post Office
Andrew Tuson, P.O. Box 30777
David Guerra
Universitat de Lleida,
Spain
City University London
Jeffrey D. Ullman,
New York, NY
10087-0777 USA
InfoLab, Stanford Postal Information
Abdelrahman Hosny
University of Connecticut,
USA
University
Moshe Y. Vardi,
XRDS (ISSN# 1528-4981)
is published quarterly in
spring, winter, summer
Rice University and fall by Association for
Andrew J. Hunsucker
Indiana University, USA
Computing Machinery,
2 Penn Plaza, Suite 701,
E DI T ORI A L S TA F F New York, NY 10121.
Vasileios Kalantzis
Director, Group Application to mail at
University of Minnesota, Publishing Periodical Postage rates
USA Scott E. Delman is paid at New York, NY
Vassilios Karakoidas
Athens University of
XRDS Managing Editor &
Senior Editor at ACM HQ
and additional mailing
offices.
Economics and Business,
Denise Doig
Greece
Maria Kechagia
Production Manager
POSTMASTER: Send
addresses change to:
XRDS: Crossroads,
Lynn D’Addessio
Athens University of Association for
Economics and Business, Art Direction Computing Machinery,
Greece Andrij Borys Associates, 2 Penn Plaza, Suite 701,
Andrij Borys,
New York, NY 10121.
Norene Kelly Mia Balaquiot
Iowa State University, USA Offering# XRDS0171
Director of Media Sales ISSN# 1528-4972 (print)
Pedro Lopes Jennifer Ruzicka ISSN# 1528-4980
Hasso Plattner Institut,
jen.ruzicka@acm.org (electronic)
Germany
Copyright Permissions
Dimitris Mitropoulos Deborah Cotton Copyright ©2016 by
Athens University of permissions@acm.org the Association for
Economics and Business, Computing Machinery, Inc.
Public Relations Permission to make digital
Greece Coordinator
Wolfgang Richter Virginia Gold
or hard copies of part of
this work for personal or
Carnegie Mellon
University, USA
classroom use is granted
without fee provided that
ACM copies are not made or
Abhineet Saxena distributed for profit or
Guru Gobind Singh Association for
Computing Machinery commercial advantage
Indraprastha University, and that copies bear
India 2 Penn Plaza,
Suite 701 this notice and the full
Olivia Simpson New York, NY citation on the first page
University of California, 10121-0701 USA or initial screen of the
San Diego, USA +1 212-869-7440 document. Copyrights for
components of this work
Gustavo Fortes Tondello C ON TA C T owned by others than
University of Waterloo, General feedback: ACM must be honored.
Canada xrds@acm.org Abstracting with credit
is permitted. To copy
Udayan Umapathi For submission otherwise, republish, post
MIT Media Lab, USA guidelines, please see on servers, or redistribute
http://xrds.acm.org/ requires prior specific
authorguidelines.cfm permission and a fee.
Permissions requests:
permissions@acm.org.
PUBLIC AT IONS BOA RD
Co-Chairs
Jack Davidson and
Joseph A. Konstan
Board Members
Ronald F. Boisvert,
Nikil Dutt, Roch Guerrin,
Carol Hutchins,
Yannis Ioannidis,
Catherine C. McGeoch,
M. Tamer Ozsu,
Mary Lou Soffa
LETTER FROM THE EDITORS

Supporting Creativity,
Expressiveness and
Complexity through
Personal Fabrication

A
s awareness of digital fabrication has grown, and access to fabrication technology
has increased, a new term has entered the cultural dialog: “personal fabrication.”
Neil Gershenfeld, the director of the Center for Bits and Atoms at MIT, is one of
the earliest proponents of personal fabrication. Gershenfeld’s description of this
field is ambitious. In FAB: The Coming Revolution on Your Desktop (Basic Books, 2005),
he describes personal fabrication as the combined use of a range of rapid prototyping
technologies—from creating custom circuit boards and programming microcontrollers,
to designing and fabricating mechani- 3-D printing, or molding and casting),
cal structures. In its most elevated and then given one week to make
UPCOMING ISSUES state, personal fabrication is not mere- something using that technique. The
ly about 3-D printing trinkets and course culminates in a multi-week
Fall 2016 keychains. Rather, it suggests how ac- project in which students combine
[September issue] cess to new technologies can enable techniques and tools from previous
individuals to create their own systems weeks to produce an interactive de-
Quantum Computing
and devices. vice or technology of his or her con-
Article deadline: June 1, 2016 The potential of personal fabrica- ception and design. When I took the
tion was conveyed to me while taking course, projects included a modular
Winter 2016 Gershenfeld’s “How To Make Almost tap-dancing robot, an LED-lit shirt
[December issue] Anything” class at MIT. In each ses- powered with thermoelectric mod-
sion, students are introduced to a ules, a self-documenting construc-
The Future of Work
new tool or prototyping technique, tion kit that built a virtual model of
Article deadline: September 1, 2016 (for example circuit board fabrication its geometry, and an open-hardware
and microcontroller programming, MP3 boombox.

XRDS • SPRING 2016 • VOL.22 • NO.3 5


INTER ACTIONS
ACM’s Interactions magazine
explores critical relationships
between people and
technology, showcasing
emerging innovations and
industry leaders from around
the world across important
applications of design thinking
and the broadening field of
interaction design.
Our readers represent a growing
community of practice that is
of increasing and vital global
importance.
To learn more about us,
visit our award-winning website
http://interactions.acm.org
Follow us on
Facebook and Twitter
To subscribe:
http://www.acm.org/subscribe
Association for
Computing Machinery
6 XRDS • SPRING 2016 • VOL.22 • NO.3
IX_XRDS_ThirdVertical_V01.indd 1
The projects from How-To-Make
demonstrate how access to digital
fabrication enables individuals to cre-
ate novel, complex, and personally
relevant forms of technology. Digital
fabrication technology is no longer
in the domain of privileged labora-
tory settings. Excitingly, more people
now have access to the technology
outside of the classroom. Over the last
15 years, CNC milling machines, 3-D
printers, laser cutters, and PCB fab-
rication have become more economi-
cally accessible for more people [1]. In
a presentation at the FabLearn Con-
ference at Stanford, Professor Mike
Eisenberg compared the 1978 Apple
II to the Makerbot 2012 Replicator 2,
predicting personal fabrication could
lead to a paradigm shift comparable to
that of personal computing.
However access to digital fabrica-
tion technology alone will not lead to
widespread and meaningful use. Rath-
er, it raises the question of how digital
fabrication will impact everyday peo-
ple’s relationship with mainstream
manufacturing. Returning to the com-
parison with personal computing, al-
though more people than ever have
access to computers, relatively few
people write their own computer pro-
grams. Instead, personal computers
frequently serve as (incredibly power-
ful) systems for accessing information
and communicating with others. This
is a valuable role for the computer to
play, but it ignores the potential of
computers to support broader forms
of creativity and innovation. Personal
fabrication faces a similar situation.
While it opens new opportunities for
more people to access and customize
physical objects made by companies, it
has the potential to be much more.
In How-To-Make, I observed an eco-
system for inspiring, supporting, and
sustaining creative and novel forms
of making. In thinking about the fac-
tors that made the course successful,
I found myself wondering how these
qualities could be translated beyond
the class, to support broader and more
diverse groups of people in designing
and making their own products.
When introducing any new technol-
ogy, one of the biggest challenges is
communicating what you can do with
it. One advantage in How-To-Make is
3/18/15 3:35 PM
Digital fabrication
is in many ways
emblematic of the
expanded role of
computer science.
computer science, architecture, and
engineering students are already fa-
miliar with traditions of design and
prototyping. These activities have
a natural fit with the capabilities of
digital fabrication. However, as we
think about personal fabrication more
broadly, it’s important to question:
Who will actually be interested in per-
sonal fabrication, and why? How can
we communicate a diversity of creative
opportunities in personal fabrication
to different people with different in-
terests? The potential is there. From
the 3-D printed puppetry in Charlie
Kaufman’s most recent animated film,
“Anomalisa” to the 3-D printed and la-
ser cut garments of fashion designer
Iris van Herpen, we’re already starting
to see strong evidence for applications
of digital fabrication beyond tradition-
al engineering prototyping. Digital
fabrication is no longer of this planet;
astronauts stationed at the Interna-
tional Space Station can create their
own replacement parts on the fly us-
ing 3-D printing. I’m convinced work-
ing to answer these pressing questions
will increase the rate and variation of
innovative applications of digital fab-
rication.
Equal to the difficulty of motivat-
ing the use of a technology for new au-
diences, is the challenge of teaching
people to use it. Personal fabrication
presents a particularly difficult learn-
ing task. People have to develop profi-
ciency with individual tools, as well as
have the knowledge and confidence to
combine multiple tools and processes.
People who are educated in engineer-
ing are often familiar with working at
different levels of abstraction. This can
make it easier to for them to learn and
apply the tools and processes inherent
to personal fabrication. Case-in-point,
in How-To-Make, the students with pri-
or experience in programing and me-
chanical or electrical engineering of-
ten fared better than those who came
from architecture, art, and design. We
can draw a lesson here when pursuing
broader forms of personal fabrication.
We must consider how people without
formal training in engineering fields
can be successful participants. This
will require the continued develop-
ment of new and more accessible fab-
rication and design tools. But it will
also require developing a variety of
learning pathways for people with dif-
ferent backgrounds and skill sets, and
scaffolding the learning process when
moving from simple workflows to com-
plex ones.
Perhaps the most important factor
in How-To-Make’s success is its com-
munity of students and its emphasis
on peer learning. For each project,
students must post online documen-
tation of how they built it, which other
students refer to in subsequent class-
es. The majority of learning in How-To-
Make is peer-to-peer; students who are
experts in CAD offer to teach others,
In its most
elevated state,
personal
fabrication
is not merely about
3-D printing trinkets
and keychains.
Rather, it suggests
how access
to new technologies
can enable
individuals to create
their own systems
and devices.
XRDS • SPRING 2016 • VOL.22 • NO.3
while students with circuit design ex-
perience create walkthroughs of how
to build and program circuit boards.
The tradition of students teaching and
supporting other students is at the core
of what drives the development of com-
plex and original projects; however,
scaling this quality of community sup-
port to a broader audience isn’t easy.
Online communities and services offer
one possibility, but they can also fall
short in supporting high-tech forms
of design and DIY. Complex online tu-
torials can be difficult for novices to
reproduce on their own [2] and accessi-
ble forms of design, like customizable
parametric models, often fail to help
people learn the skills required to cre-
ate original designs [3]. The prolifera-
tion of maker and hacker spaces, and
community events like Maker Faire,
provide more opportunities for in-
person support. Unfortunately, many
of these communities perpetuate the
diversity issues of other tech commu-
nities. For example, a 2012 survey re-
vealed only 20 percent of Maker Faire
participants are women [4]. In order to
support large and diverse communi-
ties of personal fabrication, we must
find ways to make existing digital fab-
rication communities more inclusive,
and forge connections between digital
fabrication and other “low-tech” ap-
proaches to making.
The questions and challenges I
raise might seem like challenges for
social scientists and educators, but I
argue they are also relevant questions
for the computer science community.
In many ways, computer science is
growing, from a specialized field of
study to a fundamental skill with rel-
evance in nearly all fields of human
study and expression. In his final state
of the Union, President Obama listed
helping all students learn to write com-
puter code as one of his top priorities
for 2016. As coding becomes increas-
ingly ubiquitous, the roles of computer
scientists must expand as well. More
than ever, we have a responsibility for
evaluating the social impacts of com-
putation, and for considering how new
forms of computation can appeal to
different groups of people.
Digital fabrication is in many ways
emblematic of the expanded role of
computer science. It has the potential
to change how we make things as a
society, and it also provides new op-
portunities for engaging a broader and
more diverse range of people. Com-
puter scientists play an important role
in the shaping the direction of digital
fabrication, by developing new design
software, writing systems for sharing
digital models, and collaborating with
material scientists and mechanical
engineers in developing new fabrica-
tion machines (to name a few). The
decisions we make in these tasks will
undoubtedly influence the kinds of
people who participate in personal fab-
rication in the future.
While we’re on the subject of shap-
ing the future, this letter marks the
start of a new chapter in XRDS’s histo-
ry. As of this issue, Okke Schrijvers and
myself have officially taken the reins of
the magazine. We’re extremely grate-
ful to our predecessors, Sean Follmer
and Inbal Talgam-Cohen and their
commitment to fostering the intel-
lectual depth of XRDS, and using it as
a platform to encourage engagement
and participation in computer science.
Okke and I are proud to work to con-
tinue this tradition. We’re particularly
excited that the first issue under our
tenure is digital fabrication. It spans
multiple disciplines, supports both
creative and scientific applications,
and has relevance for hobbyists, learn-
ers, and advanced practitioners. This
mirrors the way we see XRDS. Going
forward, we hope to use the magazine
to communicate and promote the mul-
titude of ways that computer science
can be meaningful to different people.
We’re eager for you to join us.
—— Jennifer Jacobs
References
[1] Mota, C. The rise of personal fabrication. In
Proceedings of the Eighth ACM conference on
Creativity and cognition (C&C ‘11). ACM, New York,
2011, 279-288.
[2] Wakkary, R. et al. Tutorial authorship and hybrid
designers: The joy (and frustration) of DIY tutorials.
In Proceedings of the 33rd Annual ACM Conference on
Human Factors in Computing Systems (CHI ’15). ACM,
New York, 2015, 609–618.
[3] Oehlberg, L. et al. Patterns of physical design remixing
in online maker communities. In Proceedings of the
33rd Annual ACM Conference on Human Factors in
Computing Systems (CHI ‘15). ACM, New York, 2015,
639–648.
[4] Maker Market Study and Media Report: An In-depth
Profile of Makers at the Forefront of Hardware
Innovation. Make. 2012.
7
begin
INBOX

XRDS STAFF UPDATES

Exciting future for
ACM XRDS [https://www.
facebook.com/acmXRDS].
Stay tuned for yet another
year of amazing issues,
updates to the magazine
to promote technical
innovation among CS
students, and increased
online presence.
—Marinka Zitnik, Facebook

Just made a post on the @

XRDS_ACM blog, all about
me: http://xrds.acm.org/
blog/2015/12/an-introduc-
tion-from-an-interaction-
designer/
—Andrew J. Hunsucker
PhD student at Indiana
University in human
computer interaction with
a focus on design pedagogy.
Semi-pro filmmaker, editor,
AE animator and VO artist.
Twitter (@AndrewJHCI)

FAVORITES FROM
THE IoT ISSUE
The Guilt Timer & so true...
(c/o @XRDS_ACM &
@PHDcomics)
—Heather Avery, PhD
Business Analytics director
for the duck, Computer
science and software
engineering PhD. #INTJ.
Twitter (@havery2)
ICYMI Vint Cerf on pros-
pects for #IoT #security
Association for Computing It was fantastic to have the —SF Bay Area ACM San Fran-
Machinery (ACM) opportunity to write this! cisco Bay Area Chapter ACM.
Public Policy Office. —Madison Capps Twitter (@SF_BayArea_ACM)
Twitter (@acmpolicy) MIT grad. Software Engineer
interested in all things brain @XRDS_ACM @TheOffici-
and vision, web development, alACM we are attempting
CONTRIBUTE TO XRDS and online education. And this now in my AI class at
Thanks @MadiCap for sometimes coffee. Poolesville High School. Art
sharing career advice Twitter (@MadiCap) or music
with #XRDS_ACM —Jack Stansbury
http://xrds.acm.org/ Software engineer and
article.cfm?aid=2845133 … #FANDOM Computer Science/Math
#softwareEngineering SFBayACM Weekly is out! Teacher at Poolesville High
Photo by Larz Photography

#privacy #interoperability #IoT #energy https://t.co/7DkPsJ7xkd School in MD. Lemelson-MIT

#standards in ACM XRDS
magazine http://xrds.acm.
org/article.cfm?aid=2845145
… @XRDS_ACM
—ACM Policy Office
—ACM XRDS Magazine Stories via @DataTorturer Excite Award Winner.
XRDS is ACM’s magazine @XRDS_ACM @vardi Twitter (@stansburyj)
for students by students. How to contact XRDS: Send a letter to the editors or other feedback by email (xrds@acm.
#XRDS_ACM. org), on Facebook by posting on our group page (http://tinyurl.com/XRDS-Facebook), via
Twitter by using #XRDS_ACM in any message, or by post to ACM Attn: XRDS, 2 Penn Plaza,
Twitter (@XRDS_ACM) Suite 701, New York, New York 10121, U.S.

8 XRDS • SPRING 2016 • VOL.22 • NO.3


INIT
Digital Fabrication: A human-machine
interface for advanced manufacturing
A
t the beginning of
the cold war, some
engineers at MIT
came up with the
idea to replace a machinist
with a computer to control
the movements of a metal
milling machine. Because
of the high demand for air-
planes, and the progress
made in aerodynamics,
complex curves needed to
be fabricated. Machinists
skilled enough to fabri-
cate those shapes were in
short supply. Computers
have no trouble calculating
axes’ positions on complex
curves, and so combining
computers and fabrication
machines seemed like a
great idea. The first forms
of computer numerical con-
The focus
of this issue
is therefore
as much about
the newest
developments
in digital
fabrication,
as it is about
the yearnings
we have for
the future.
XRDS • SPRING 2016 • VOL.22 • NO.3
BioBots’ 3-D printer combines a special ink with
living cells to build 3-D miniature human organs and
living tissues, which can be used for personalized
therapy and as a replacement for animal testing.
trol (CNC) were developed. though. Thus a bubble was of machine control and de-
Some 20 years later, addi- born that wasn’t entirely sign projects.
tive manufacturing using in line with the realities of In fact, making new
stereolithography was also digital fabrication. machines for making has
invented, and the first 3-D Digital fabrication prom- become a field of its own,
printers were developed us- ised a revolution in manu- and avenues of inquiry
ing that technology. These facturing. Shapes and around machine build-
computer-controlled fab- forms that were impossible ing are opening up. It’s as
rication methods and ma- to make traditionally for though using machines to
chines are now referred to low-volume without high- make has resulted in more
as “digital fabrication.” cost were now possible, as making of machines that
Digital fabrication uses were unprecedented cre- make. Making machines
codes that define opera- ation and customization by that make is part of what
tions, for example, design end-users and laypeople. the focus of digital fabri-
files that instruct machines To some extent, digital fab- cation research is. New,
on how the parts can be
made. The hope was digi-
tal fabrication would make
advanced fabrication as ac-
cessible as digital compu-
tation had made comput-
ers. Manipulating these
codes was, after all, some-
thing that anyone could
do; any computer could be
used to edit the codes that
defined the objects, and
sharing codes could be as
simple as sending an email
attachment. Digital fabri-
cation, and especially 3-D
rication has delivered on
those promises, but not by
exciting digital fabrication
tools harnessing precision
printing, sounded like it surrounding us with “Star and creativity are being re-
could be the science fiction
future we had been prom-
Trek” replicators. Although
we didn’t experience the
leased to complement the
commodity digital fabri-
ised! However, in reality, “Diamond Age”, improve- cation tools, which are al-
the industrial machines
that were being used for
ments in digital fabrication
did give us commodity 3-D
ready on lab benches and
studio spaces everywhere.
digital fabrication were dif- printers, laser cutters, and Digital fabrication re-
ficult to operate, expensive,
and not that flexible. 3-D
CNC mills. People have be-
gun to chip away at the old,
search therefore has a wide
reach. How are things cur-
printed parts were small, clunky human-machine in- rently being made on the
brittle, and mostly made of
plastic. That didn’t stop the
terfaces, customize robot
arms for fabrication, and
shop floor? What are new
processes that are being de-
hype around 3-D printing develop many other kinds veloped to print new materi-
9

als and assemble new struc-
tures? What are the tools
soon to be released to the
public? What are the tools
that are barely viable dem-
os in laboratories? What
are the file formats, design
software, and control soft-
ware being used? How is all
this 3-D printing and digi-
tal fabrication going to be
environmentally sustain-
able? The focus of this issue
is therefore as much about
the newest developments in
digital fabrication, as it is
about the yearnings we have
for the future.
Manufacturing used to
encompass many differ-
ent professions, from CAD/
CAM designers to machin-
ists. Now those things can
all be done by the same per-
son using digital fabrica-
tion tools. What does that
mean for the possibilities
of personal-scale digital
fabrication? And what does
personal-scale digital fabri-
cation mean for individual
creativity and digital fabri-
cation? Ilan Moyer explores
these questions in “Per-
sonal Fabrication Tools:
From automated machines
to augmented tools,” and
Amit Zoran presents a “A
Manifest for Digital Imper-
fection” for personal cre-
ation.
Technology is not always
easy to change. How do you
make it so a user can incor-
porate later interactions
into the tools? Stelian Coros
10
NASA is financing research to feed astronauts in
space using 3-D printing; food will be transformed
from a powdered form, which has a shelf life of 30
years, into customized and nutritious meals.
explores characters of in-
teractive objects in “Creat-
ing Animated Characters
for the Physical World”;
while Andrew Spielberg,
James McCann, Alanson
Sample, Scott Hudson, and
Jennifer Mankoff explore
the possibilities of RFID to
make objects more interac-
tive in “Building a Toolkit
for Fabricating Interactive
Objects.” Valkyrie Savage
explores augmenting 3-D
printed objects with elec-
tronics for interactivity in
“3-D Printing Interactive
Objects.”
Right now, 3-D printers
mostly print in rigid ma-
terials or elastomers. They
don’t print soft materials
like yarn or textiles. How
can we create machines
that print with the same
Making new
machines for
making has
become a field
of its own,
and avenues
of inquiry
around
machine
building are
opening up.
range of materials we find
in the physical world?
Huaishu Peng, Scott Hud-
son, Jennifer Mankoff, and
James McCann show one
example in “Soft Printing
with Fabric.”
Zahner Architectural
Metals has an R&D depart-
ment where engineers such
as James Coleman, Craig
Long, Andrew Manto, and
Trygve Wastvedt figure
out how to make buildings
with thousands of unique
parts. Jesse Louis-Rosen-
berg is half of the compu-
tational design studio Ner-
vous System. Zahner works
on huge buildings and
Nervous works on intricate
jewelry; yet they both have
the same problems: Cur-
rent file formats and work
flows for digital fabrica-
tion are not well matched
to their tasks. How can we
design better infrastruc-
ture for digital fabrication,
either in voxel file formats
or in completely reconfigu-
rable machine workshops?
They outline their head-
aches and suggestions in
“Lots of Parts, Lots of For-
mats, Lots of Headache”
and “Drowning in Triangle
Soup: The quest for a better
3-D printing file format”
respectively.
But even with the cur-
rent file formats, digital
fabrication is useful in
fields where custom manu-
facturing is the norm rath-
er than the exception. Erin
XRDS • SPRING 2016 • VOL.22 • NO.3
Buehler outlines the link
between digital fabrication
and assistive technology in
“Fabrication Lends a Hand:
Creating Custom Assistive
Technology.”
Finally, what is the long-
term vision of digital fabri-
cation? Will it help us move
to space? Will it help us
settle on Mars? In “ Island
Three Revisited: O’Neill’s
cylinders and digital mate-
rials,” Daniel Cellucci and
Kenneth C. Cheung outline
what digital fabrication
might mean in the far fu-
ture. Closer to home, Dave
Rejeski has some concrete
suggestions for keeping 3-D
printing sustainable in “3-D
Printing: Green or not?”
The diversity one finds
in how people use digital
fabrication machines will
be paralleled by the diver-
sity of digital fabrication
machines that are created.
In this issue, we have col-
lected just a few examples
of where we can see that di-
versity today. In the future,
we hope for an expressive
and powerful collection
of tools, processes, and
formats. If the kinks of
working across computers,
machines, data, and op-
erators can be addressed,
we think this will happen
pretty soon. It might even
be revolutionary.
—Stefanie
— Mueller
and Nadya Peek,
Issue Editors
 In 2015, the number of official Fab Labs
worldwide doubled from 256 to 549.

ADVICE

How to Deal with Negative Critique

O
ne of the first things a gradu-
ate student must learn is how
to deal with critique. While
assessment exists at all levels
of higher education, graduate students
are expected to be professional and
able to deal with harsh critique.
Critique can come from lots of
places in graduate life, but the most
common and challenging critics will
be conference and journal peer re-
viewers. They are notorious for being
overly harsh, and sometimes may have
an agenda; there have been incidents
where reviewers demand additional
references to a particular author (who
might be the reviewer themselves), or
otherwise challenge solid research be-
cause it doesn’t necessarily align with
their own work.
The peer review system is typi- thing to do when receiving a peer re- tee the editor will agree with you, or
cally double-blind, meaning the au- view that seems unfair is to get a sec- reverse the decision on your paper,
thor doesn’t know who the reviewer ond opinion. It’s sometimes difficult but it’s important the editor is aware
is and the reviewer doesn’t know to separate yourself from your work, of your concerns as this will affect the
who the author is. This is important, especially given the large amount of reputation of their publication.
because it removes biases and pro- effort that goes into a conference or Receiving a rejection and harsh cri-
tects reviewers from further retribu- journal paper. Show the review to your tique on a paper you’ve spent months
tion. The paper committee chair for advisor. A good advisor will be able to working on can be one of the most
a conference, or the journal editor, tell you if the critique is actually unfair brutal experiences a grad student can
knows who both the author(s) and or if it’s simply harsh. go through. But the most important
the reviewers are. Therefore ensuring Occasionally, a review will not just thing to remember is this: There’s al-
reviewers don’t receive papers from be harsh, but might appear to be re- ways another place to send your paper.
authors from their own institution. If viewed by someone who doesn’t un- Some reviewers will hate your paper,
the editor or chair is doing their job, derstand your field, or asks you to some will love it, and some will feel
your paper will be reviewed by expe- make alterations for reasons that don’t it just isn’t right for the venue you’ve
rienced members in your field. These seem valid. If you come across a review submitted it to. Some reviewers will
reviewers will be able to spot flaws in like this, again getting the opinion of be kind and supportive in rejection,
your research and arguments easily. your advisor is a good place to start. If and others will be savage. But their
Image by Ollyy / shutterstock.com

Remember, when a paper is published
it becomes a part of the literature of
the field. The job of a reviewer is to
validate whether the paper that has
been submitted is a significant contri-
bution to the field.
Just because a review is harsh, that
doesn’t mean it’s also unfair. The first
they agree the review is problematic,
your next move is to inform the jour-
nal editor or conference paper chair.
These people are in place to make the
final decision on your paper for their
respective publication. Explain your
objection clearly, and give examples
from the review. There’s no guaran-
goals should be the same: to ensure
the highest quality research gets pub-
lished in their field.
When you receive your reviews,
take the critique, find the important
points, discard the rest, edit your
work, and resubmit.
—Andy
— J. Hunsucker

XRDS • SPRING 2016 • VOL.22 • NO.3 11

begin
UPDATES
RIT SIGCHI
Democratizing digitization,
campus outreach, and more.
P
hilo Farnsworth invented
the television in 1928, but it
took more than 20 years for
commercial TV to take off. In
1946, only 6,000 U.S. households had
a set, within five years 12 million did.
Decades later, in 1983, Chuck Hall
brought us stereolithography, more
commonly known as 3-D printing. It
took even longer for this technology to
go mainstream. Although 3-D printing
has become a ubiquitous term, and one
can buy a printer at Home Depot or Sta-
ples, arguably it is as likely to be used
to make a novelty plastic toy as well as a
life-saving medical device.
The two different technologies,
with different objectives and invent-
ed in different eras, are nevertheless
RIT SIGCHI At a Glance
School:
Rochester Institute of Technology
Chapter Name: RIT SIGCHI
Location: Rochester, NY (USA)
Website:
https://www.facebook.com/RITHCI
Date Established: November 5, 2013
Officers:
Sabarinathan Masilamani
Chair
Jeremiah Parry-Hill
Treasurer/Secretary
Amanda Yung
Communications
Ajantha Vijayasekharan
Membership and Events
Current Total Membership: 45
Contact: rit.sigchi@gmail.com
12
Establishing the U.S. Postal Service as
3-D printing hubs could streamline the
fast delivery of printed goods and boost
package revenue up to 20% per year.
uniquely similar: When invented, each
was ungainly, slow, expensive, hard
to use, and for many years, not par-
ticularly useful or practical. Dr Dan
Ashbrook, director of the Rochester
Institute of Technology’s (RIT) Future
Everyday Technology Research Lab
(FETlab), wants to change that para-
digm by using, ironically, 3-D printing.
At a talk hosted by the school’s ACM
student chapter in November, Ash-
brook claimed the goal and promise
of “democratizing digital fabrication”
was to expedite the process of moving
an invention from the impractical to
the practical. To further that goal, the
RIT chapter will play a major role—
nearly half of his FETlab research as-
sistants are student members.
Established in November 2013 with
faculty advisor Deborah Gears and a
handful of students, the RIT special
interest group for computer-human in-
teraction—RIT SIGCHI—is the official
ACM student chapter at RIT. Situated
330 miles northwest of New York City,
near Lake Ontario and the Canadian
border, it has grown to 45 members in
its three-year existence.
Sponsoring talks and hosting infor-
mational social gatherings comprised
the bulk of the chapter’s outreach ef-
forts in 2015. At its fall semester kick-
off event, the group held a pizza-fueled
social gathering to discuss recent
trends in HCI, as well as view a num-
ber of videos on the topic. “In the life
of the graduate student, which is nor-
mally focused on balancing studies
with life, simply providing a social out-
let around these topics means a lot in
terms of building a community around
HCI,” said Jeremiah Parry-Hill, who
holds the dual role of treasurer and
secretary for the chapter. The chapter
held a similar event last April, which
featured HCI-centric films and videos.
Also that same month, the group host-
ed Matt Huenerfauth who gave a talk
on methods for evaluating the effec-
tiveness of technologies for conveying
American Sign Language.
Social media is another important
aspect of the RIT SIGCHI community.
The chapter uses its Facebook page to
keep members and other interested
parties informed of upcoming pro-
fessional opportunities, free or low-
cost research material, and other an-
nouncements. In addition, for those
unable to attend chapter-hosted pre-
sentations, SIGCHI established a You-
Tube channel1 to view both Ashbrook
and Huenerfauth’s talks.
The chapter aims to expand its
reach to those beyond its current mem-
bership. A robust crowd of 20 attended
Ashbrook’s talk. About half were RIT
SIGCHI, while the rest were faculty
members and other interested parties.
“We have a ‘big-tent’ philosophy,” Par-
ry-Hill explained, “where we want to
get our message to anyone interested
in human-computer interaction—first
to members, then the school, then to
local universities, and ultimately to
the western New York region. That will
take time.”
“Part of it is logistics issues,” add-
ed Amanda Yung, the organization’s
communications chair. “We’ve had
1 https://www.youtube.com/channel/UCsBOsq_
pRqxl2hVGwmeYWvA
XRDS • SPRING 2016 • VOL.22 • NO.3
 $3 The amount you need
to fabricate a functional
3-D printed gun.
ACM
ACM Conference
Conference
Proceedings
Proceedings
Now
Now Available via
Available via
Print-on-Demand!
Print-on-Demand!

Did you know that you can

now order many popular
ACM conference proceedings
via print-on-demand?

Institutions, libraries and

individuals can choose
from more than 100 titles
on a continually updated
list through Amazon, Barnes
& Noble, Baker & Taylor,
Ingram and NACSCORP:
CHI, KDD, Multimedia,
SIGIR, SIGCOMM, SIGCSE,
SIGMOD/PODS,
and many more.
Photograph by Julia Prisaznuk. © RIT Golisano College of Computing and Information Sciences.

For available titles and

ordering info, visit:
librarians.acm.org/pod
Tirumalavan “Balaji” Iyengar, a member of RIT SIGCHI, presents a storyboard to his
classmates. He and his team built a physical Internet-quality metering device that
drips water to reflect the speed of the upstream and downstream Internet connec-
tion, as well as display an LED bargraph of the same information.

some events at other universities lo-
cally, for example, in 2013. But getting
people to and from one campus to an-
other for a one hour presentation can
be a barrier. Videoconferencing, more
for logistical reasons, hasn’t been a
viable option.” The chapter intends
to explore options for increasing its
range in the coming months.
The RIT SIGCHI chapter also ex-
pects to continue its work supporting
the democratization of digital fabrica-
tion in the school’s FETlab. “The chap-
ter will have a hands-on workshop with
Professor Ashbrook in the spring se-
mester,” said Parry-Hill.
Back in November, Ashbrook opened
his presentation quoting techno/sci-
ence-fiction author William Gibson,
“The future is already here, it’s just not
very evenly distributed.” With the help
of RIT SIGCHI, students are learning,
and working, to change that.
—David
— Byrd

XRDS • SPRING 2016 • VOL.22 • NO.3 13

begin

At TED2015, Carbon3D’s Joseph DeSim-

one suggested the possibility of 3-D
printing a tooth in less than 10 minutes,
while you sit in the waiting room.

CAREERS

How to Get Hired at a Startup

as a Software Developer
Editor’s Note: Voxel8’s Jack Minardi
walks us through the resume,
the interview, and to the only advice
you need to get hired at a start-up:
Starting your own.
—Bryan Knowles

A
ccording to software engineer
and writer Joel Spolsky, man-
agers should hire people who
are smart and get things done.
For many applicants, this is done
by demonstrating prior work history.
However, those just coming out of
school may need to be more creative
in how they showcase those talents. In
this article, I will share some pointers
on what you can do to put these traits
forward in your job applications.

SHOW OFF
One good way to show you are pas-
sionate about software is to program
for the fun of it. This can take many
forms—from contributing to an open
source project that you find useful,
to starting a new project from the
ground up. While I think the easiest
way to find motivation to start your
own project is to dream up some-
thing that scratches your own itch. If
you are having trouble thinking up a
project there are communities that
are full of good ideas waiting for mo-
tivated programmers to step up, such
focused on ideas that he simply en-
joyed doing.
If you do end up starting your own
project, you need to put it online so
that others can see and use it. Both
Github and Bitbucket are good places
to host your code. Moreover, these
sites have the added benefit of teach-
ing you good distributed source con-
trol tools, like git. Once you’ve put
some of your code online, it is a good
idea to put a link to your Github or
Bitbucket profile on your resume.
Many companies ask to see examples
of your previous work, and this is an
different aspects of software develop-
ment, from programming to working
on a team. Project-based courses also
give you a chance to hone your leader-
ship skills, which are valuable within
a startup.
SEE THINGS THROUGH
Demonstrating you can get things
done is a little more difficult than dem-
onstrating you are smart. Getting good
marks is one way, but the interview
process itself also provides a critical
opportunity to show what you can do.
Be prepared to talk to potential em-
Image by Dotshock / shutterstock.com

as reddit.com/r/somebodymakethis.
Doing technical work purely for
fun can sometime lead to incredible
results. In The Pleasure of Finding
Things Out, physicist Richard Feyn-
man said the inspiration for his No-
bel prize-winning work came when
he stopped trying to search for seri-
ous problems to tackle and instead
easy way to showcase it.
The classes you take in college are
another good way to demonstrate
how smart you are. Take challenging
classes that advance your knowledge
and allow you to practice skills you
will use in the real world. To that end,
I recommend project-based courses,
as these provide experience in many
ployers about projects you’ve driven to
completion and the key things you’ve
learned along the way. I would much
rather see someone with one complet-
ed project than 10 half finished ones,
and seeing good documentation and a
test suite is always a good sign. During
the interview, your goal is to show you
have an understanding of what it takes

14 XRDS • SPRING 2016 • VOL.22 • NO.3

 2025 In less than 10 years, McKinsey
estimates the economic
impact of 3-D printing to be as
large as half trillion dollars.

MILESTONES

Additive
Manufacturing
to finish projects. When I interview Additive manufacturing, also known as 3-D
people, I look for evidence of this. printing, is a technique to produce three-
Demonstrate that you know every dimensional objects by printing successive
completed engineering project has layering of material. This field promises an
tradeoffs. It is a good idea to discuss exciting future with many plausible applications
what tradeoffs you’ve made in past for everyday life including food, medicine, and
projects. (This is a common inter- technology. Let’s examine a few milestones from
view question!) Sometimes you need the past 30-plus years in the development of 3-D
to choose a technically inferior solu- printers.
tion for time or budget reasons. When
talking about tradeoffs, I am remind-
ed of a famous quote from The Soul
of a New Machine: “Not everything
1980 Hideo Kodama of NMRI (Japan)
invents the first rapid prototyping
system. It uses photopolymers, which harden
worth doing is worth doing right.” when exposed to UV light.
Not everything in a shipped prod-
uct will be perfect. A good engineer
knows which aspects must be ideal
and which can be merely adequate.
1989 S. Scott and Lisa Crump invent and
patent fused deposition modeling
(FDM), a technique used in many commercial
and open-source 3-D printers today. FDM uses a
BE PATIENT nozzle to dispense molten plastic or metal on top
A more appropriate title for this ar- of a controllable movable table.
ticle might have been, “How to Get
Hired at Voxel8 as a Software Devel-
oper.” Although every company is dif-
ferent, the advice here is sufficiently
1999 Researchers at the Wake Forest
Institute for Regenerative Medicine
develop synthetic, growable human bladders
general enough to be applicable to that use a combination of 3-D printed scaffolds
most startups. Remember, startups and human cells from patients.
are especially picky about who they
hire, since each employee can make
or break a team. Most good startups
know this, and frankly you shouldn’t
2005 The RepRap project begins as
an open-source initiative to develop
an inexpensive 3-D printer. The project released
work for the ones that don’t. four official printer models from 2007 to 2010.
Nonetheless, to get hired at a start-
up takes patience. The jobs you want
will reject you, and the only guaran-
teed way to get hired at a startup is
2014 Local Motors unveils Strati,
the first electric 3-D printed car.
The car took 44 hours to print and reaches
to start one yourself. That may be the a speed of approximately 50 miles per hour.
only piece of advice you need.
—Jay
— Patel
Biography
Jack Minardi is the co-founder and software lead at
Voxel8. He graduated from Vanderbilt University with a
B.E. in electrical engineering and worked as a scientific
software developer at Enthought. He then joined the Lewis
Lab at Harvard University as a research fellow, where he
led efforts in precision machine automation and control.

XRDS • SPRING 2016 • VOL.22 • NO.3 15

begin

The XRDS blog highlights a range of topics from conference coverage, to security
and privacy, to CS theory. Selected blog posts, edited for print, are featured in
every issue. Please visit xrds.acm.org/blog to read each post in its entirety. If you
are interested in joining as a student blogger, please contact us.

BLOGS

important detail. While VR completely blocks out the

outside world, AR creates an overlay on the existing world.
This could be used to keep meta-information about the
environment available, such as directions to a location
or weather information. Recently Microsoft unveiled
the HoloLens, which will allow players to turn any room
into a gaming space. At E3 2015, Microsoft demoed their
HoloLens with Minecraft, showing how players could play
in multiplayer worlds, and even change aspects of the world
with voice commands (see opening image).
But AR tech is still in its infancy. The HoloLens is a huge
leap forward, but it’s clear VR tech is where most of the
developer resources are going. Although in this article,
we’ll be talking mostly about VR, AR isn’t going away. Thus
it is important to think about what will VR and AR look like
in 5-10 years?
Let’s now look at how current experiences are being
adapted to VR, how successful those experiences are, and
specific deficiencies that designers will need to improve
before the technology can reach its full potential.

AUGMENTED REALITY THEATER

While working on my master’s degree in human computer
interaction at Indiana University, I along with my colleagues
Exploring Virtual Reality— (Corrie Colombero, Pui Mo, and Monet Rouse) looked at the
world of virtual and augmented reality, and came up with a
Are We There Yet? design that we thought might fill some gaps in the current
By Andrew J. Hunsucker experience. The result was the Augmented Reality Theater,
or the ART (see Figure 2).
Virtual reality (VR) is the next hot technology. The Oculus The ART was designed for 5-10 years into the future. It
Rift, a virtual reality headset that allows you to see 3-D used AR glasses and tactile experiences to enhance the
worlds, is almost ready for consumers and developers are rac- movie theater experience.
ing to come up with novel uses for it. Most of the ideas lean This design was eventually presented at SIGGRAPH
toward the entertainment realm. Video games in particular
have been a fertile ground for experimentation in the field of Figure 2. A mockup of what users of the Augmented Reality
VR. Adapting existing video games to VR is a great first step Theater (ART) might see.
to better understanding this technology, and what it can of-
fer. But the technology could theoretically offer a lot more.
VR promises the ability to put humans into worlds
and situations that would otherwise be impossible or
too dangerous to experience. It does this by completely
immersing the user’s vision with a video headset. This
headset tracks their head movement, so the user can look
around naturally, adding to the immersion. Noise canceling
headphones are often used to immerse the player’s hearing
as well. The only thing the user can see or hear is what the
designer wants them to experience. This can be incredibly
powerful in the hands of a creative and talented designer.
This is different from augmented reality (AR) in an

16 XRDS • SPRING 2016 • VOL.22 • NO.3

2014 and we learned a lot about the world of VR and AR.
Working on this project gave me a sense of how advanced
the technology in these worlds was, but from the perspective
of a designer, it also showed me how much work needed to
be done to improve these experiences.
The point of the ART was to create a novel movie theater
experience. However, as we learned more about the world
of AR and VR, it became clear that there were some obvious
deficiencies in the current experience.
EXAMINING VIRTUAL REALITY
Because of the fact that VR replaces at least two of the users
senses completely, we can judge how successful a VR experi-
ence is by how well it immerses the user in this new world.
While a user can have their vision and hearing immersed
in this new world, as soon as they attempt to stand up and
move around, or speak, or touch something in the real world
that doesn’t match what they are seeing and hearing, the
illusion is broken.
Designers have been attempting to solve these problems
to create complete, full-body immersion. This would allow
the user to not just look around the world, but to have their
full-body movements replicated in the virtual world; allow
the user to feel and manipulate the objects in the virtual
world; and allow the characters in the virtual world react to
their speech and movements.
The ideal experience looks
something like the Holodeck from
“Star Trek”.1
In the show, simulations are
so advanced and realistic to be
dangerous, even deadly. We’re a long
way from that experience—assuming
it’s even possible—but there are a
few problems with current VR that if solved, would greatly
enhance the user experience in VR. In adapting existing and
familiar experiences into VR, the three current problems
I see are gesture controls, free movement, and tactile
experiences.
GESTURE CONTROLS
Gesture controls are a significant challenge in all comput-
ing right now. Current control mechanisms like the mouse
and keyboard have been in use since the dawn of desktop
computing, but when the user isn’t chained to a screen,
these control mechanisms don’t make sense any more.
While it is possible to control VR experiences with a mouse
1 https://www.youtube.com/watch?v=oZwtVz7z0wM
XRDS • SPRING 2016 • VOL.22 • NO.3
Figure 3. A player experiences the PrioVR system by leaning
over. His avatar in the game responds by leaning over as well.
and keyboard, for a fully immersive experience, designers
are looking for new ways to give the user full control.
Let’s take a common example that many VR experiences
are trying to adapt: the first person shooter (FPS). FPS
games allow the player to use a variety of guns to fight their
way through a scenario, shooting enemies throughout the
game. The genre is so named because the player sees the
environment from a first-person perspective, seeing only the
gun in their field of view.
PrioVR (http://www.priovr.com/) is a technology that
utilizes an Oculus Rift headset and a full body suit to track
the player movements. This suit translates their movements
into movements for their avatar in the game (see Figure 3).
Their video demonstration2 shows the player picking
up, aiming, and firing weapons; leaning around corners;
and even performing melee attacks, both punches and
kicks. However, looking at the video, it doesn’t look entirely
accurate or natural. And a player using this system in their
home would need a lot of space to perform the movements
that the player in the video is performing. With their vision
completely covered, an errant movement could easily destroy
equipment in the home, or injure the player.
Another control device called Sixense (http://sixense.
com/wireless) uses a modular system that involves handheld
controllers with additional tracking points for the limbs
and torso. These additional points appear to be optional.
The handheld controllers track very precise movement of
the hands, and their video3 suggests players will be able to
naturally pick up objects in the game world (see Figure 4).
This controller appears to track gestural movements very
2 https://www.youtube.com/watch?v=q72iErtvhNc
3 https://www.youtube.com/watch?v=Ux89C3YhPVc
17
begin
Using 3-D printing, Nervous
Systems has created a plastic
dress that can flow like fabric and
fold/unfold like an origami.
BLOGS
Figure 4. The Sixense STEM system translates user move-
ments into gameplay movements.
precisely, and helps translate those movements into in-
game actions. One of the demonstrations involves the player
holding lightsabers and deflecting laser blasts from a drone.4
Sixense’s attempts to solve the problem of using the
device in a smaller space. The player can set up sensors
around their play area, and if they get too close to the edge,
the game world will show them a red laser grid, warning
them that they are getting too close to the edge of the play
area. I see this as a good compromise between the limited
space that most home players would have to use this system
in, and giving players the confidence that they would need to
play this type of game.
Regardless of whether it works perfectly at this point, it
is clear we have the technology that will be required to track
player motions and translate them into an experience. While
I wouldn’t call it a solved problem, there is definitely thinking
being done, and with more development, this part of the
experience will be much improved.
A much bigger problem is free movement.
FREE MOVEMENT
When immersing a user’s sight and hearing in a virtual
world that allows them to look around, tracking their head
movements precisely, it’s not surprising that the user might
expect to be able to walk around in this virtual world, and
interact with objects more fully. However, that’s not the
experience that VR normally offers.
The reasons for this are pretty simple. The virtual world
doesn’t need to, and shouldn’t translate directly to the real
world. The entire point of VR is to put users into a world
4 https://www.youtube.com/watch?v=2lY3XI0zDWw
18
that they couldn’t normally visit. Recreating a player’s
living room while they sit in their living room is probably
not an experience most people would be excited about.
Several companies are attempting to solve this
problem. Cyberith (http://cyberith.com/) and Virtuix
(http://www.virtuix.com/) are both developing devices
that hold the user in place, while allowing their feet free
movement through the use of a low-friction surface. This
allows the user to use their legs freely, while not allowing
them to move around and run into walls or equipment
nearby (see Figure 5).
When combined with a VR headset, along with some kind
of controller (in their promotional material, both companies
use a gun-shaped controller for FPS games), these
technologies could successfully simulate free movement in a
virtual world (see Figure 6).
However, these systems both require a lot of extra
equipment and space. It’s not clear if I could strap myself
into one of these systems, or if I’d need assistance.
And any tech this new is sure to be expensive. The
Cyberith system is available for pre-order for $1249, as of
this writing. For a new laptop, that’s a great deal, but for a
gaming peripheral, it’s quite expensive. The Virtuix Omni
is available for pre-order for $699 as of this writing, which is
much cheaper comparably, but still incredibly expensive.
Both of these systems also include custom-sized
components, like harnesses. The Virtuix Omni requires
special shoes. This means different-sized people would
have to purchase extra equipment in order to make use of
the system.
But these are not entirely out of the reach of someone
who has the means to purchase a gaming PC and other
Figure 5. On the left, a user of the Cyberith system. On the
right, their visual within the game.
XRDS • SPRING 2016 • VOL.22 • NO.3
 3-D printed prosthetic hands can be
downloaded online for free, 3-D printed
in 20 hours, and assembled in 2-4 hours,
costing as little as $20.
Figure 6. A player experiences the game Skyrim through the
use of the Virtuix Omni.
entertainment devices. And they both appear to work with
a host of popular games, so there’s no reason to think
that future games would ignore their technologies as they
become more common.
Another company called Zero Latency (https://
zerolatencyvr.com/) is taking a different approach. Rather
than hold the player in place, they are simply giving them a
much larger area to roam in.
This company puts players into a large warehouse
space, and has created a custom game to be played with
VR headsets and gun controllers. Patrons can play six
player co-op to fight off zombie hordes with completely
free movement.
Giving players open space to move around in adds greatly
to the immersion, but creates some safety concerns. The
company claims they have in-game safety features that will
keep you from walking into walls, and offer an hour of time
in the game for $88 per person.
However, this experience only exists in one place
(Melbourne, Australia), and there is only one experience
to be had. It’s hard to fault them for either of these issues
though, as the technology is so new, and their solution to
some of the problems I’ve considered is quite novel.
While other companies are trying to make these
experiences smaller and more compact, Zero Latency
have recognized one of the big advantages of the VR
experience is its size. Rather than hold people back,
they’ve given them more space to roam. It will be
interesting to see if they can expand to other locations,
XRDS • SPRING 2016 • VOL.22 • NO.3
and other experiences. As it stands now, it looks like
they’ve got a really great start.
But all of these companies have only partially solved the
last problem I want to talk about: tactile experiences.
TACTILE EXPERIENCES
As humans move through the world, all of their senses are con-
stantly engaged. They can see the world around them and hear
the sounds of the birds chirping and people talking. They can
feel the breeze blowing past them, a drop of rain on the top of
their head, a hand on their shoulder, or the transition between
the sidewalk and a dirt path under their feet as they walk.
Sight and sound are powerful experiences, but recreating
the sense of touch in a virtual world is a very complex
problem. Our entire bodies are capable of pulling in
information through the skin. Temperature, moisture, and
pressure are all types of information that we can understand
without seeing or hearing anything.
This has largely been ignored in the virtual world. I
don’t believe it’s because developers and designers feel it’s
unimportant, but more likely because it’s a difficult problem
without a clear solution. Some gaming systems we’ve already
mentioned give players a gun-shaped controller which is
perfectly usable during FPS games.
This solution works great for FPS games, and since many
companies are using these games as an introduction for
their technology, it’s a great first step to enhance immersion,
and a quick win.
But as VR experiences expand, finding a universal
solution for tactile experiences will become more and more
challenging. Every physical prop that replicates something
in the virtual world reduces the flexibility of that scenario.
Imagine a virtual car prototype. A designer shows the
user the car they are designing via a VR headset to get
reactions. The user reaches out to touch the steering wheel,
but there’s nothing there. So the designer places an analog
for the steering wheel in place to increase engagement in the
prototype. The user now has a steering wheel to touch, but
now reaches out to touch the radio, which isn’t there, or the
turn signal knob, or the door release.
If the designer continues adding physical pieces to this
experience, eventually they’ve just built a physical model,
which is what they were trying to avoid by building the
virtual prototype. But without a tactile experience, they lose
one dimension of the experience they’re testing.
When researching the ART, my team and I worked
on discovering all sorts of interesting tactile technology
in development. Like REVEL and Aireal from Disney
Research. REVEL (http://www.disneyresearch.com/project/
19
begin
BLOGS
revel-programming-the-sense-of-touch/) is a wearable
technology that changes the way users experience surfaces.
For instance, if the user reached out to touch a smooth
surface, but the VR experience was showing a stone
surface, this technology could give the user the experience
of touching stone.
Aireal (http://www.disneyresearch.com/project/aireal/)
uses small puffs of air to simulate impacts on the user. Their
demonstration uses soccer balls bouncing off
the users hands, but any kind of impact could theoretically
be simulated.
We don’t live in a Star Trek world where we can use
force fields and other high tech tricks to simulate objects.
So why not design experiences that don’t require the user
to touch anything?
This is one of the reasons that Birdly (http://somniacs.co/)
is so successful. (see Figure 7).
Birdly is a full body VR experience that allows the user to
experience flying. The system has the user lie face down in a
rig that allows them to flap their arms like wings. They can
also lean left, right, forward and back in order to control the
experience. They are strapped into the system and it requires
outside assistance to get into and out of. In addition, a fan is
placed near their head to simulate the wind blowing against
them. I got to see Birdly at SIGGRAPH 2014, and while I
didn’t get a chance to experience it myself, everyone I talked
to thought it was very successful.
To me, its success is based largely on embracing the
constraints of the medium. Rather than trying to fit an
existing experience into the VR headset, the creators of
Figure 7. A demonstration of the Birdly system. The user is
strapped into the machine with a VR headset, and a fan blows
air on them to simulate wind.
20
Birdly created their own experience that matched what they
could accomplish with the headset, and worked from there.
CONCLUSION
I see Birdly as the most successful VR experience that cur-
rently exists. Rather than try to recreate reality, it creates
its own reality. In doing so, it only has to follow the rules it
creates for itself. This allows for a much richer experience.
The user isn’t spending time wondering why they can’t
reach out and touch something, they are totally immersed
in the experience.
If your user would be disappointed in the inability to use
their hands in your experience, then turn them into wings. If
your users would be confused that they couldn’t walk around
this virtual world, let them fly.
VR is only going to become bigger and better over the
next 5-10 years while AR and other related technologies
are developing. The thinking on these topics will only
expand. However, I would like to see less thinking on how
the technology works, and more thinking on what gestures
humans will find comfortable, and how it will actually feel as
they are using it.
Right now, one of the biggest issues I see with VR is too
much energy is being spent on improving the technology
itself, and not enough is being spent on understanding
the user experience. I come from a design background,
and I think VR tech is at the point where the headset
and audio are good enough for us to start thinking
about how humans actually want to use it. Most of the
experiences I’ve discussed are in their infancy. They’re
demonstrations that exist in order to show users the
possibilities of the technology.
Much like any new medium, it begins by adapting
existing material. When film first arrived, they looked much
like plays, with a still shot viewing a scene that actors would
work in. When television first arrived, they borrowed many
of their programming ideas from radio and film. So it’s no
surprise that VR experiences are adapting video games as
the technology becomes viable.
But an experience like Birdly gives us a peek into what
is possible with a VR system. If we want this tech to be
successful, we have to start embracing the constraints of the
medium and come up with new experiences, rather than
simply adapt existing experiences into this new technology.
I look forward to the future.
Biography
Andrew J. Hunsucker is a Ph.D. student at Indiana University studying Design Pedagogy
and storytelling. He is currently studying the HoloLens from Microsoft and hopes to better
understand how users can interact with their environment in virtual reality spaces.
XRDS • SPRING 2016 • VOL.22 • NO.3
 Promoting Computing Ethics

The Pledge of the Computing Professional is an organization for emphasizing

ethical and moral behavior to students graduating from computing‐related
degree programs. The Pledge entails a voluntary rite-of‐passage ceremony
featuring an Oath written by computing professionals that complements and
supports the ACM Code of Ethics. This rite‐of‐passage ceremony is conducted
by Nodes (that is, chapters) currently at over 30 academic institutions throughout
the United States and, soon, internationally.

Establish a Node at your institution:

www.computing‐professional.org
Endorsed by ACM SIGCAS, ACM COPE, and the Order of the Engineer

A personal walk down the

computer industry road.
BY AN EYEWITNESS.
Smarter Than Their Machines: Oral Histories
of the Pioneers of Interactive Computing is
based on oral histories archived at the Charles
Babbage Institute, University of Minnesota.
These oral histories contain important messages
for our leaders of today, at all levels, including
that government, industry, and academia can
accomplish great things when working together in
an effective way.
feature

A Manifest for
Digital Imperfection
Artistic style is an important aspect for creative practice.
However giving away some computational control over
digital design and fabrication is necessary in order
to engage designers in a higher-risk practice that enhances
attention, creative decision making, and product ownership.

By Amit Zoran
DOI: 10.1145/2893491

S
“ cience is an endless search for truth. Any representation of reality we develop
can be only partial. There is no finality, sometimes no single best representation.
There is only deeper understanding, more revealing and enveloping
representations.” —  Carl R. Woese [1]

The very basis of the current computational paradigm is the assumption that analytic
representation is key in developing practical systems to solve technical tasks. The foundations
of computer science are rooted in discrete (and Boolean) mathematics, where binaric
machineries and computational systems define whether an input—or function—belongs to
a given set, whether a problem is While the computational sciences to design style—whether generated by
solvable or not, whether a solution and analytic representations undoubt- computer-aided design (CAD) or com-
to a given problem can be applied edly leverage very powerful tools to puter-aided manufacturing (CAM)—
to a different problem, and what is understand the world and develop and suggest a revision of the way we
the optimal way to solve such ques- advantageous technologies, they are conceptualize computers for creative
tions. Similarly, in many other fields not free from weaknesses and limita- outcomes.
of technology, engineers are using tions. While some negative phenom-
continuous mathematical tools to un- ena of the Information Age can easily THE PROBLEM OF STYLE
derstand, analyze, predict, and simu- be observed (cyberbullying, early-age IN COMPUTATIONAL DESIGN
late real-world phenomena; to decide exposure to sexual content, and so In many fields of computer science,
whether a signal represents a symbol; on), other issues such as the diffi- researchers successfully identify indi-
and to build reliable machinery that culty of representing cultural values vidual users by features extracted from
delivers expected results. Further- mathematically are less visible. Many their recorded performance (such as
more, modern (quantum) physics scholars have already articulated the gait tracking and recognition). More-
contributes statistical observations risks digital agency presents to com- over, computational methods have
of uncertain reality, suggesting addi- puter-aided society [2]. In this article, contributed to the evaluation of cre-
tional non-deterministic methods to I highlight an additional limitation ative style within computer graphics
interpret the world and conceptualize in the current computational design (CG), where researchers have devel-
new quantum technologies. and fabrication paradigm in relation oped tools to extract the identifying

22 XRDS • SPRING 2016 • VOL.22 • NO.3

Photo Credit TK

XRDS • SPRING 2016 • VOL.22 • NO.3 23

feature

characteristics of artists’ styles. For

Figure 1. The Hadza hunters in Tanzania mark their arrows with example, in Image and Video-Based
unique personal patterns. Artistic Stylisation, CG researchers
list methods for “Stroke-Based Paint-
erly Rendering” (chapter one), “Artis-
tic Stylization by Nonlinear Filtering”
(chapter five), “Artistic Rendering of
Portraits” (chapter 12), and more [3].
This body of work considers style as
a static quality. Changes in personal
style over time have not been mapped,
and the relationship between style and
time-varying motoric skills (the rela-
tionship between subjective creative
intention and technical abilities) have
not been studied.
But what is creative style? In arche-
ology and anthropology, the study of
material culture has motivated de-
Figure 2. Violin back plates made from maple wood. Left: A back plate produced tailed evaluations of variability in
with a digital milling machine, rendering a uniformed pattern on the wood. Right: human techniques and methods for
A back plate made by the author using hand tools, showing a complex pattern that making artifacts. Starting with the
reflects the variability of techniques being used. early 20th century work of the German-
American anthropologist Franz Boas
[4], the study of artistic style within the
humanities is an important part of re-
search into material culture. Kroeber
has described style as “a self-consistent
way of behaving... selected out from
among alternatively possible ways...
selective with reference to values” [5].
Work by Wiessner suggests the style
of craftspeople in traditional practices
reveals social information and expres-
sions of personal identity [6]. In anthro-
pology, it seems to be a consensus that
personal style is unique—the charac-
teristic signatures of individual mak-
ers identify them among other makers,
yet the signals communicating this
variability are culturally dependent
(see Figure. 1). These maker-signatures
are not always evident to the observer,
and the relationship between style and
skill is not entirely clear.
On one hand, creative style has been
the subject of many studies. By creative
style, I mean the formal variations in
artwork or design that transmit infor-
mation about personal and social iden-
tity. This is a variation on Wiessner’s
definition of style as a “formal varia-
tion in material culture that transmits
information about personal and social
identity” [6]. However, individual style
in carrying out creative and expressive
motoric tasks (using a free-hand prac-
tice, usually by a skilled person) has
not been studied in depth with quan-

24 XRDS • SPRING 2016 • VOL.22 • NO.3

titative methods within computer sci-
ence. Most work in computer science
seeks a working algorithm to achieve
acceptable results when implanting
formal variations in artwork or design
(such as brush strokes qualities and
patterns, color distribution, and so on)
outside their original creative context.
Examples include products meant to
stimulate the eye with signals that re-
mind us of a well-known artist, or that
fit current standards of aesthetics [7].
For instance, the painting style of
Vincent Van Gogh can easily be iden-
tified; the now-popular artist used
distinct expressive techniques in his
work. However, the collective memory
of these historical works is biased to-
ward iconic examples or even toward
some cognitive morphing between
them. Thus, a Photoshop (or Google
DeepDream) filter designed to stylize
your photos as if they were painted by
Van Gogh may satisfy users, yet bear
little resemblance to the portraits Van
Gogh would have created today had he
lived to the age of 163. An artist’s style
is never frozen in time; it reflects the
context of the work. Imagine, for ex-
ample, what would have motivated Van
Gogh to dedicate a work to you, and
how would this initial motivation in-
fluence the preliminary conditions of
his work?
The dangers of the current CG trend
are hidden deep in the intangible qual-
ities of the creative practice, not all of
which are easy to define. Perfection,
homogeneity, and uniformity of shape
and texture do not, by themselves, car-
ry cultural signals—these signals are
produced and consumed by people. A
uniform texture may demonstrate ac-
curacy, but does not carry a great deal
of information, while a complex tex-
ture can reveal much more about the
designers and their cultural context
(see Figure. 2). Similarly, a perfect mod-
el of a virtual idea does not tell us any-
thing about the fabrication process,
but an imperfect reproduction embeds
a story within the artifact. However,
the economic influence of computa-
tional machinery convinces us that
computers need to play an important
role in creating cultural artifacts for
design and art. Recently, researchers
have invested in developing a variety of
software agents that “takes input from
XRDS • SPRING 2016 • VOL.22 • NO.3
Instead of aiming
to simulate style or
imitate it, we should
use this complex
quality to close the
gap between digital
designers and
digitally fabricated
artifacts.
designers, then ‘evolves’ new designs
on its own,” since “designing products
[traditionally] is costly and time-con-
suming” [8].
To summarize, research in both
computer science and humanities
falls short in its contributions to the
development of new, style-oriented
expressive design tools. Although CG
researchers study neither style nor
aesthetics from a humanistic perspec-
tive, they develop tools to transplant a
certain distribution of visual elements
outside of their original context in or-
der to generate a desired emotional
reaction. Archaeologists and anthro-
pologists, on the other hand, study
style in its cultural context, but do not
develop new tools for creative produc-
tion. Nor have they studied in depth
the dynamic evolution of individual
style over time.
STYLE, TIME, AND MEANING
People change over time: Their mo-
toric skills change, their intentions
change, their styles change, and their
cultural contexts change. A creative
practice ties abilities with intentions,
but what are these abilities and inten-
tions? Is the metaphysical separation
between abilities and intentions, skill
and style, and matter and form useful
when we study creativity? If so, can we
study creative intention without devel-
oping a psychoanalytic model of the
artists and their cultural contexts?
Obviously we can develop a con-
vincing digital style generator, which
we can use to gain a profit in specific
economic contexts. Nevertheless, such
developments do not reveal impor-
tant questions about style—they only
generate a simulation, by implanting
out-of context signals to synthesize a
reaction meant for a different context.
Alternatively, we can choose to rely on
Kantian philosophy, reject the duality
between abilities and intentions, and
focus on the empirical interaction be-
tween human and matter. However,
even that direction won’t make the
analytic comprehension of style or its
analytic representation more feasible.
Visual design is the art of convey-
ing symbolic meaning within an arti-
fact, rather than solving an optimiza-
tion problem. While some aspects of
machine and interface design can be
defined as engineering problems, de-
sign is not a branch of engineering. It
deals with underlying cultural signals
of aesthetics and meaning, with the
aim of emotionally engaging people
with the subject of the design work. A
good interface design guarantees us-
ers will succeed in completing a task,
but it will also impact their emotional
reaction to the product.
The cognitive and affective aspects
of design are interwoven [9]. Some of
the dependencies of this relationship
fit well into computational sciences,
while other aspects do not. Still, recent
trends in digital design and fabrication
introduce computers, with their ana-
lytic problem-solving manner, to the
whole spectrum of the design process.
Thus re-imagining design as an engi-
neering challenge and implementing
continuous control on the whole cre-
ative chain—from autonomous design
generators to digitally controlled fabri-
cation machines.
Today, the vision of big-data beau-
tifiers (styling products based on well-
observed retail trends), which rely on a
low degree of personal preference, in-
fluence design paradigms and impact
research in both industry and acade-
my. But it is a problematic vision. While
digital technologies have altered the
design practice, allowing for powerful
CAD and CAM tools, they overshadow
some design qualities rooted in cultur-
al meaning and material context.
Hence, I argue for a revision of this
trend, advocating for digital imperfec-
tion in computational design practices
as a way to conjure a struggle between
creative skill and personal style—a
25
feature
Figure 3. A watercolor painting of a chameleon using an augmented airbrush [11],
which allows novices to experience the manual art of spray painting.
Unlike an automatic printer, here the virtual simulation guides the physical process,
allowing for a new experience with a singular physical style.
struggle that will contribute to an un-
predictable yet meaningful product.
This is not a Marxist critique on mass
manufacturing and the alienation it
creates between labor, society, and
human nature. It is a manifest for im-
perfection and openness: An imperfec-
tion that can engage the designer in an
ongoing performative search for bal-
ance between the controllable and the
uncontrollable in design and nature. A
search to balance forces and interests,
rather than simulating aesthetics that
will generate predictable product ac-
ceptance. A manifest for preventing
Figure 4. The FreeD in action.
26
the virtual agency from over control-
ling all aspects of physical matter.
FROM IMPERFECTION
TO CREATIVITY
The question of style in computational
design can be considered as a ques-
tion of agency. Artists and design-
ers develop their personal styles in a
melting pot, where skill, aesthetics,
environmental influences, and design
objectives mix together and influ-
ence each other. When we outsource
responsibility for style to an automa-
ton—by implementing a simulation
of the styles of skilled artists out of
their original context, by expecting
the computer to beautify work based
on programmable aesthetic criteria,
or by using a digital machine to build
artifacts—we are giving up on human
engagement and decision making. We
are, in fact, giving up on part of the hu-
man creative spectrum.
Nevertheless, while computational
agency dominates modern production
lines and design tools, there is still a
space to encourage the development
of digital design tools that engage a
broad range of users in creative prac-
tices. Although many digital tools aim
to ease design and fabrication, I ar-
gue openness, unpredictability, and
imperfection in creative tasks help to
engage the designer in an intense ex-
perience that has a higher potential
to generate subjectivity and meaning,
and support artistic expressive state-
ments and ownership (see Figure 3).
One example of such an interaction
is the FreeD device [10], a hand-held
digital milling device that is moni-
tored by a computer, while preserving
the maker’s freedom to manipulate
the work in many creative ways (see
Figure 4). Relying on a pre-designed
3-D model, with the FreeD the com-
puter springs into action only when the
milling bit risks the object’s integrity,
preventing damage by slowing down
the spindle. The rest of the time, it al-
lows complete gestural freedom.
A user study of the FreeD device re-
veals how synergetic cooperation be-
tween human and machine preserves
the expressiveness of manual practice
[10]. This quality of the hybrid terri-
tory evolves into design personaliza-
tion. Moreover, the study shows signs
of correlation between the early man-
ual style of the participants and their
style while using the digital device
(see Figure 5). Based on interviews
with the participants, who are all de-
signers, the authors report this corre-
lation is gained by engagement in the
practice of making while facing chal-
lenges that appear during the work, or
ideas evolving while using a carving
tool. As such, it could not have been
realized before the fabrication proc-
ess. This form of involvement allows
performance of personal style within
a digitally monitored fabrication task
XRDS • SPRING 2016 • VOL.22 • NO.3

Figure 5. Five users execute the same design using the FreeD.
Figure 6. A castle model fabricated using the FreeD, and the tool path of the device
being used by the author in making this castle. We tend to develop technologies
focusing on the final results rather the process itself, yet the process contains
a rich information about subjective creative decisions.
that helps compensate for the user’s
lack of carving knowledge without au-
tomating the process.
Zoran et al. suggest a hybrid interac-
tive system will be beneficial for open-
ended processes, allowing makers to
define the amount of computational
control they use. Beginners may need
guidance to simply complete the task
at hand, while developing their tech-
niques as part of the investigation. On
the other hand, skilled makers may
require higher-level control, allowing
the computer to reproduce their skills
or manually seeking different objec-
tives, such as introducing random
qualities to the process. The image of
human–computer synergy is subjec-
tive and should be open-ended and
variable if it is to support real creative
engagement.
The FreeD enables users to interpret
and modify a virtual model during fab-
rication, keeping the user’s subjective
XRDS • SPRING 2016 • VOL.22 • NO.3
tool path as a signature embedded in
the texture of the physical artifact. Be-
cause the FreeD allows design manip-
ulation to be integrated within a tan-
gible carving experience, the nature of
this work more closely resembles the
process of traditional craft than other
forms of digital fabrication, while still
providing digital risk management
and quality assurance. By introducing
traditional approaches to the digital
making of artifacts, Zoran et al. hope
this intimate collaboration between
people and computers will pave the
path for a new type of interaction.
Building upon prior art such as the
FreeD, I hypothesize personal style in
expressive manual tasks can signal
rich information (see Figure 6). More-
over, this time-varying information
depends on the task itself, while also
revealing identifying characteristics
of the observed subject. I believe these
identifying characteristics can be care-
fully extracted from records of creative
performance, using properly tuned
models. More specifically, we should
study which characteristics of creative
style depend on elements of the cre-
ative performance, such as the mor-
phology of the task, its social context,
the mental model, and the user’s expe-
rience, and which time-varying charac-
teristics are unique to the user and ap-
pear in all of her or his work. Instead of
aiming to simulate style or imitate it,
we should use this complex quality to
close the gap between digital design-
ers and digitally fabricated artifacts.
And also aim to enhance expressivity
and unpredictability, which can arise
in less controllable or “perfect” virtual
design environments.
References
[1] Woese, C. R. A new biology for a new century.
Microbiology and Molecular Biology Reviews 68, 2
(2004), 173–186. DOI: 10.1128/MMBR.68.2.173–186.
[2] Turkle, S. Alone Together: Why We Expect More from
Technology and Less from Each Other. Basic Books,
New York, 2011.
[3] Rosin, P., and John, C. (Eds.). Image and Video-Based
Artistic Stylisation. Springer-Verlag London, London,
2013.
[4] Boas, F. Primitive Art. Reprint. Dover Publications,
Mineola, NY, 2010.
[5] Kroeber, A. L. The Nature of Culture. University of
Chicago Press, Chicago, 1952.
[6] Wiessner, P. Style and social information in Kalahari
San projectile points. American Antiquity 48, 2
(1983), 253–276.
[7] Pavlidis, T. and Van Wyk, C. J. An automatic beautifier
for drawings and illustrations. In Proceedings of the
12th Annual Conference on Computer Graphics and
Interactive Techniques (SIGGRAPH ’85). ACM, New
York, 1985.
[8] Kinkead, G. Software designs products by simulating
evolution. MIT Technology Review. November 7, 2014.
http://www.technologyreview.com/news/532126/
software-designs-products-by-simulating-evolution
[9] Ingold, T. The textility of making. Cambridge Journal
of Economics 34, 1 (2010), 91–102.
[10] Zoran, A., Shilkrot, R., Nanyakkara, S., and Paradiso,
J. A. The hybrid artisans: A case study in smart tools.
ACM Transactions on Computer-Human Interaction
(TOCHI) 21, 3 (2014).
[11] Shilkrot, R., Maes, P., Paradiso, J. A., and Zoran, A.
Augmented airbrush for computer aided painting
(CAP). ACM Trans. Graph. (TOG) 34, 2 (March 2015),
Article 19.
Biography
Dr. Amit Zoran is a senior lecturer at the School of
Engineering and Computer Science at the The Hebrew
University of Jerusalem. He holds a Ph.D. and a M.S. in
media arts and science from the MIT Media Lab, a M.Des. in
product design from Bezalel, and a B.Sc. in communication
system engineering from Ben-Gurion University, Israel.
In his work, Zoran studies human-computer interaction,
design, and craft, exploring the divergent realms of
emerging computational design technologies and
traditional hand-hewn skills.
© 2016 ACM 1528-4972/16/03 $15.00
27
feature

Personal Fabrication:
From automated
machines to
augmented tools
Fully automated digital fabrication tools are the darling of the personal fabrication
movement, but they may not be the best format for harnessing digital fabrication for
personal use. Instead we should be developing tools that work cooperatively with users
to augment natural abilities rather than eliminate human involvement altogether.

By Ilan Moyer
DOI: 10.1145/2896905

D
igital fabrication is an exciting set of technologies that use a computerized workflow
to first design and then precisely create physical objects. These tools have existed as
the workhorses of industry soon after their invention in the 1950s, but only in the
past decade have they been miniaturized, cost-reduced, and simplified to the point
that they are now accessible to the general public. Inexpensive (or free) computer-aided
design (CAD) tools, such as SketchUp and Autodesk Fusion360, make it easier for anyone to
design nearly anything. Small computer numerically controlled (CNC) milling machines and
3-D printers are popping up at an astonishing rate in makerspaces, hackerspaces, schools,
museums, and garages across the world. While this trend promises
to vastly expand the abilities of the in- must traverse to fully integrate into a lery guns, and numerically controlled
dividual to create, there is still a long new context of “personal fabrication” milling machines were invented to
way to go before digital fabrication tru- has a precedent: personal computing. better produce complex airplane com-
ly finds a widespread home on peoples’ In his book Fab: The Coming Revolution ponents. Computers were then adopt-
desks and in their garages. The best on Your Desktop—from Personal Com- ed by industry for tackling laborious
solution to the challenge of democra- puters to Personal Fabrication, Neil Ger- computational tasks like payroll, and
tizing digital design and fabrication is shenfeld draws an apt parallel between digital fabrication was used for tasks
likely not through the incremental cost digital fabrication and the rise of the like high-speed high-tolerance manu-
reduction and simplification of indus- personal computer in the ‘70s and ‘80s. facturing. It was not long after that
trial tools, but instead demands a fun- Both technologies were first developed
damental re-examination of how we for military purposes around the time
use computers to help us make things. of WWII; early computers were used to Figure 1. The Shaper Origin computer-
The path that digital fabrication calculate fire control tables for artil- assisted cutting tool.

28 XRDS • SPRING 2016 • VOL.22 • NO.3

Photo Credit TK

XRDS • SPRING 2016 • VOL.22 • NO.3 29

feature
CACM_TACCESS_one-third_page_vertical:Layout 1 6/9/09 1:04 PM Page 1
the lockstep held between these sister
technologies was broken. While the
computer began to rapidly evolve into a
consumer product, digital fabrication
remained exclusively a tool of industry.
By the early 1970s, the mainframe
ACM was shrunken to desktop propor-
tions. Although the technology had
been made human-scale, the human-
Transactions on machine interface remained an 80
character-wide glowing cathode-ray
Accessible tube. The early adopters of personal
computers were enthusiastic and pas-
Computing sionate enough to subject themselves to
an interface whose designers were more
likely interested in maintaining a status
quo than improving user experience. It
wasn’t until the widespread introduc-
tion of the Macintosh in 1984, with its
graphical user interface and desktop
metaphor, that personal computation
had finally taken a form accessible and
intuitive to the average person. Person-
al computation was enabled by a fun-
damental re-invention of how people
interact with an existing technology.
Just as computers transitioned
from the military-industrial complex
to the realm of personal use, the same
is happening now with digital fabrica-
tion tools almost 50 years later. In the
words of Gershenfeld, “personal fabri-
cation is the ‘killer app’ of digital fabri-
cation.” There are many benefits to this
◆ ◆ ◆ ◆ ◆ new style of digitally enabled personal
fabrication. One is the abstraction of
This quarterly publication is a
skill. You can shape complex metal or
quarterly journal that publishes plastic parts with your digital hand
refereed articles addressing issues
of computing as it impacts the
lives of people with disabilities.
The journal will be of particular
interest to SIGACCESS members
Instead of forcing
and delegates to its affiliated
conference (i.e., ASSETS), as well
the user to adopt a
as other international accessibility
conferences. technology-centric
◆ ◆ ◆ ◆ ◆ system designed
www.acm.org/taccess for industry, digital
www.acm.org/subscribe fabrication should
fit itself into the
natural way people
are already using
tools to make things.
30
simply by drawing the shape in a CAD
program. Related to that is repeatabil-
ity. Just as a mathematical expression
will evaluate the same value on any
computer system, ideally a 3-D model
fed to a digital fabrication tool should
result in identical outputs, whether
you are sitting in a garage in the U.S.
or a makerspace in Norway. This is es-
sential to the third benefit, which is the
ease of digitally sharing plans. Placed
within the context of the Internet and
accessible CAD tools, these attributes
enable someone with little skill to
go online, download a plan, perhaps
make a few changes, and press “print.”
However, along with these benefits
come some distinct challenges. In
spite of its new personal fabrication
audience, the process of using digi-
tal fabrication has remained largely
identical to the practices of industry.
The end-to-end workflow starts by de-
signing an object in a CAD program,
followed by the use of computer-aided
manufacturing software to create in-
structions for an automated machine
to follow, and concludes with the set-
up and operation of a CNC tool. In an
industrial mass-production context,
segmenting the use of a tool into these
distinct steps is logical. Each domain
requires specialized training and a
large corpus of task-specific knowl-
edge and experience. Indeed efficiency
via specialization has been a dominant
strategy of industry since at least the
days of Henry Ford. The technology
historian David Noble even went so far
as to argue the segmented specializa-
tion of digital fabrication—enforced in
part through deliberately complicated
machine interfaces—was a tactic of
management to maintain hierarchy
in their organizations. Yet as digital
fabrication now makes the leap from
industry to personal fabrication, it is
important we question whether there
might be a better way of interacting
with this powerful set of tools.
Digital fabrication is stuck in a rut
equivalent to personal computers of
the 1970s. Just like the awkward period
preceding the introduction of the Ma-
cintosh, today’s personal digital fabri-
cation tools have been miniaturized to
fit your desk, but little has changed in
the way they are expected to be used.
You still need to learn how to use CAD
XRDS • SPRING 2016 • VOL.22 • NO.3
software, often times learn how to gen-
erate tool paths, and then learn the
nuances of setting up and operating a
digital fabrication machine.
All of this is required to generate
any output. In contrast, the operation
of a hand tool occurs on a much more
analog scale. As you acquire skill the
output improves, but even a first-time
user can create a first-order approxi-
mation of their goal. The success of the
Macintosh was largely in its approach
of making the technology of compu-
tation intuitive to its user, rather than
insisting that the user bend to the
technology. The graphical user inter-
face and desktop metaphor were key
to this, as were the subtle hints perva-
sive to the user experience (often found
in its iconography) that suggested the
machine was a bit “human.” Digital
fabrication could serve to learn from
the early success of the Macintosh.
Instead of forcing the user to adopt a
technology-centric system designed
for industry, digital fabrication should
fit itself into the natural way people are
already using tools to make things.
There are a few examples of this
starting to happen already. One recent-
ly announced tool, Glowforge’s 3-D la-
ser printer, is probably the first laser
cutter to be designed from the perspec-
tive of the personal fabrication cus-
tomer rather than clinging to the prec-
edent set by existing tools. An onboard
camera makes it easy to locate digital
shapes to be cut on a piece of material,
and a user can even draw with a pen di-
rectly on the material. Glowforge then
uses computer vision to trace and cut
along hand-drawn lines. The camera
is also used for automatically snap-
ping certain designs to known objects
(e.g., the back of a MacBook), detecting
material types by reading barcodes
printed with invisible ink, and cutting
out a large shape in multiple steps by
finding where prior cuts ended.
Another example is the work be-
ing done by my own company Shaper.
We are designing a computer-assisted
handheld cutting tool called Origin
(see Figure 1). It enables users to cut
complex shapes from materials like
wood, plastic, and light metals with
the accuracy and at the scale of a
large CNC tool, but entirely freehand.
Origin works using a combination of
XRDS • SPRING 2016 • VOL.22 • NO.3
Digital fabrication
is stuck in a rut
equivalent to
personal computers
of the 1970s.
“tool GPS,” which can locate the tool
on the material within thousandths of
an inch, and a small automated stage
that is constantly adjusting the posi-
tion of the cutting blade to account
for user error. The experience of using
Origin is equal parts hand tool and
video game—you move the tool along
the work-piece, doing your best to fol-
low the outline of the digital shape
you want to cut on the onboard screen.
Meanwhile the tool makes minor cor-
rections to your position to ensure your
cut ends up exactly as intended.
The driving idea behind the tool
is not about its underlying technolo-
gies of computer vision and motion
control. Rather, Origin is challenging
a pervasive assumption that the
re-contextualization of digital fabri-
cation for personal fabrication must
steadily approach the single-push-
button paradigm of the “Star Trek”
replicator. Fully automated digital fab-
rication tools make perfect sense in a
mass-production environment, where
speed and absolute repeatability are
paramount and operator intervention
is expensive. There is little pressure
for industrial automated tools to be
easy to use, because the cost of pay-
ing a highly skilled technician to set
up the tool is quickly amortized over
the volume of a production run. These
considerations are quite different than
what should matter in a digital fabri-
cation tool intended for personal use.
In this new environment, ease of use
is the most important consideration.
Accuracy is important, but not all-
important. Flexibility is essential, be-
cause the average person can’t afford a
new machine to do a specific job. And
the trade-offs of time and money weigh
less heavily at home, so it’s OK if the
“cycle time” to produce an object is a
bit longer. There’s also a long history
of pride in hand-craftsmanship that,
while still sometimes present in the
abstract sense in mass production, is a
very real thing in personal fabrication.
Origin takes a step toward restoring
the craft practice of concurrent think-
ing and making, which were intention-
ally divorced by industry, while still
offering the many benefits of repeat-
ability and communication afforded by
digital fabrication.
We are steadily approaching a cross-
roads in the nature of our relationship
with our tools. One path is leading us
to fully automated universal fabrica-
tion machines—like the replicator—as
household appliances. In this future,
the physical act of making will be as
simple as pressing a button on a com-
puter screen, similar to pressing “print”
to send a photo to an inkjet printer. All
of our skills and craft become interac-
tions with a computer screen. This is
the terminal stop on the vector current-
ly established by traditional industrial
digital fabrication technologies. The
alternative path is leading us to an era
of personal fabrication when intelligent
tools, such as Origin and Glowforge,
will enhance the innate physical abili-
ties of their users rather than attempt-
ing to eliminate their involvement in
the fabrication process altogether. Of
course these two potential futures are
not necessarily mutually exclusive. In
any event, it is difficult to make a quali-
tative judgment of which might be pref-
erable. Fully relinquishing our ability
to make things with our own hands in
favor of push-button convenience could
be the price of a future of highly custom-
ized abundance for all; or it may be just
enough rope to hang ourselves. But re-
gardless of which future is more appeal-
ing, it is clear digital fabrication needs
to take a different—and I would argue
more human—form if it is to be widely
adopted by individuals as the empower-
ing tool it has the potential to be.
Biography
Ilan Moyer is a co-founder and mechanical engineer at
Shaper, where he is helping to build the future of power
tools. Moyer has been interested in creating tools for
personal fabrication since writing his MIT undergrad thesis
on the topic of Rapid Prototyping of Rapid Prototyping
Machines in 2008.
© 2016 Copyright held by Owner(s)/Author(s).
Publication rights licensed to ACM.
1528-4972/16/03 $15.00
31
feature

Creating Animated
Characters for
the Physical World
Making the design and production of animated,
mechanical characters accessible to the public.

By Stelian Coros
DOI: 10.1145/2893493

T
he Maker movement is revolutionizing the accessibility and appeal of engineering.
Key to the growing popularity and success of this do-it-yourself culture is the
emergence of easy-to-use, computer-aided design (CAD) software—such as Google’s
SketchUp and Autodesk’s 123D suite—that allows the general public to generate
content for 3-D printing. However, to a larger extent, such user-friendly CAD tools focus
on supporting the design of static objects—essentially sculptures. At Carnegie Mellon’s
Robotics Institute, our long-term goal is to create robotic devices that can sense, move, and
interact with the world.
As robots enter our daily lives, they phology. To date, more than 180 million dressed. First, engineering principles
have the potential to fundamentally customized animated characters have and standards have to be encoded in
improve the way we work, learn and been created with this software by non- software such that computing power
play. With the aid of specialized CAD expert users from around the world. can be leveraged in assisting users
tools, and leveraging the 3-D printing with tasks that are too difficult or too
revolution, we believe the general pub- COMPUTATIONAL DESIGN tedious. Second, user interfaces need
lic will be able to create robots that are OF ANIMATED MECHANICAL to provide ample room for creativity,
specifically designed according to their CHARACTERS while allowing users to specify, at a
individual needs and preferences. Our The first challenge in creating a ro- high-level, what it is they wish to cre-
ambition is to make the design of per- botic creature is to design its mechani- ate. To begin investigating these chal-
sonalized robotic creatures equally cal structure. This task is currently lenges, we developed a computational
powerful, accessible, and appealing. restricted to skilled and experienced design system for mechanical autom-
In this quest, we are inspired by engineers, who, given sufficient time ata—simple robots composed of rigid
the ease of use, power and flexibility and resources, can design complex ro- mechanical components such as gears
of the Spore Creature Creator system bots, mechanical toys, or animatronic and linkage structures. The motion of
described by Hecker et al. [1]. Spore figures capable of producing life-like such mechanical characters are driv-
Creature Creator enables the design of motions. However to make this design en by a motor or a manually-operated
virtual characters whose motions are process accessible to the general pub- crank (a device used to input rotational
automatically adapted to their mor- lic, two main challenges need to be ad- motion into a mechanical system) [2].

32 XRDS • SPRING 2016 • VOL.22 • NO.3

 Photo Credit
Image reprinted
TK from SIGGRAPH Asia 2015. DOI: http://doi.acm.org/10.1145/2816795.2818137

XRDS • SPRING 2016 • VOL.22 • NO.3

33
feature

Our design system assumes ac-

Figure 1. An overview of our computational design system for mechanical cess to a library of mechanisms that
characters. (a) An input character is provided, and the relative position of the gear are used as the building blocks. Each
box is specified; (b) the desired motion of the character is specified by sketching type of assembly is parameterized,
motion curves; (c) optimized driving mechanisms are generated by our framework; and, in an offline step, our system
(d) gear trains are created to connect all the driving mechanisms to one actuator; performs a sparse sampling of their
(e) a support structure is generated; and (f) the resulting mechanical character is parameter spaces. This precomputa-
3D-printed. tion step results in a database of rep-
resentative motions that each type of
(a) (b) mechanical assembly can perform.
After this offline step, our computa-
tional design system allows casual
users to interactively create animat-
ed mechanical characters.
As illustrated in Figure 1, the user
crafts the motion of an input character
(see Figure 1a) by sketching curves that
indicate, for example, the path its feet
should move along (see Figure 1b). For
each input sketch provided by the user,
our system generates an optimized me-
chanical assembly—a collection of gears
and rigid links that reproduce as closely
as possible the motion envisioned by
the user. The process of generating the
optimized mechanical assemblies lever-
(c) (d)
ages the database of mechanisms that
was precomputed in the offline step.
This precomputed database, in essence,
plays the role of an experienced engi-
neer. The optimized mechanical assem-
blies that are generated are connected
to the input character through pin joints
(see Figure 1c), driving its motions in a
way that is as close as possible to what
the user specified. This iterative design
process continues until the user speci-
fies all aspects of the motion they wish
to create (see Figure 1d).
Each mechanism generated by our
design system requires a source of ac-
tuation in order to generate its motion.
(e) (f) In line with traditional mechanical
automata designs, however, we wish
to generate mechanical characters
that require just one motor or hand-
driven crank. Our design tool, there-
fore, generates a collection of inter-
connected gears that transmit motion
from the input actuator to the rest of
the mechanical character (see Figure
1d). Once all the required mechanical
parts are in place, our software system
generates support structures that hold
all the components together (see Fig-
ure 1e), and the design is then ready to
be 3-D printed (see Figure 1f).
As an alternative to using a set library
of mechanism types, we also developed

34 XRDS • SPRING 2016 • VOL.22 • NO.3

a design tool that allows non-experts to
create, from scratch, complex linkage Figure 2. The character design process illustrated on the Satyr’s leg: (a) input
structures, which are networks of rigid animation, (b-e) interactive motor replacement, (f) after global optimization,(g)
bars connected to each other through and after linkage shaping.
mechanical joints. Linkage structures
integrate form and function in a seam-
less manner, as their motions are the
result of complex interactions between
the bars and the constraints imposed
by the joints.
The problem of designing linkage (a) (b) (c) (d) (e) (f) (g)
structures has fascinated mathemati-
cians for centuries. Kempe’s 1876 Uni-
versality Theorem, for instance, states desired motion, or worse, mechanical results and experiments. Even non-
planar curves defined by polynomial locking of the assembly, barring motion expert users can quickly create aes-
equations can be traced out by link- altogether. Our computational design thetically pleasing animated charac-
ages, leading to the famous “there is a system circumvents these difficulties ters, as illustrated in Figure 3.
linkage that signs your name” quote [3]. by allowing the user to quickly explore
The proof of the theorem relies on a few multiple valid configurations for the in- COMPUTATIONAL DESIGN
elementary types of linkages that are termediate linkage structures. OF DEFORMABLE CHARACTERS
connected to each other to create com- In order to make this possible, we While some virtual characters are rig-
plex linkage systems. Unfortunately, developed a fast method to compute idly articulated, others are inherently
the theory also predicts the number of approximately optimal parameters for soft and deformable. Their motions,
such elementary linkages that must be new mechanical linkages, provided rather than being defined by joint
composed together quickly becomes an analysis of singularities that can angles, are continuously propagated
intractable in practice. arise in linkage systems, and proposed throughout their entire structure, as
To strike a balance between motion a method to optimize the motion of visualized in Figure 4. When creating
quality and overall aesthetic appeal, linkage-based characters and their physical embodiments of such flexible,
our computational system puts the robustness against fabrication errors. virtual characters, the ways in which
user in charge of overseeing the design With our computational system, the they deform should be taken into ac-
process. The process starts with an an- design process becomes tractable and count and reproduced as closely as
imated virtual character—for example enjoyable; as indicated by our early possible. By analyzing the equations
a running bull. We assume this input
animated character can be decom- Figure 3. The mechanical Taurus. The linkage structure of its fore leg (left), and
posed into several complexes (one for the completed prototypes (right).
each leg, for example) that act within a
predefined plane.
Our design tool enables the user to
create linkage structures whose motion
matches the animation of each com-
plex. We begin by representing each
animated joint in the complex as a vir-
tual motor. In principle, the motion of
each joint could be realized in a physi-
cal system through the use of servos.
However, the resulting design would
likely be too expensive. Therefore our
system successively replaces virtual
joint motors with new rigid links that
mechanically couple the motions of
different parts of the structure, as seen
in Figure 2. Manually performing this
operation can be a very tedious and
frustrating task. The space of possible
designs is large, highly nonlinear, and
presents both discrete and continuous
parameters. This combination of chal-
lenges leads to frequent design pitfalls
including poor reproduction of the

XRDS • SPRING 2016 • VOL.22 • NO.3 35

feature

Figure 4. The steps to designing deformable characters. An input character and target shapes are taken as input; a small
number of actuators and their initial placements are determined; the positions of the actuators and the internal distribution of
soft and stiff materials are optimized; and the physical prototype is fabricated.

of motion of elastic objects, it becomes
clear that controlling their deforma-
tion behavior can be achieved in sev-
eral different ways: adapting their rest
configuration (the shape they would
take in the absence of any external
forces), changing the material proper-
ties (e.g. stiffness) governing the rela-
tion between forces and deformations,
or explicitly applying external forces
that are appropriately modulated. In
simulation, the process of automati-
cally adapting the rest state of deform-
able objects in order to control their
motions works well [4]. However, the
technology that allows fabricated ob-
jects to significantly change their rest
configuration in a controllable man-
ner is currently unavailable.
In recent work, we introduced a
method that automates the process of
Figure 5. A snapshot of the design interface. Left: The design viewport with the
footfall pattern graph. Right: the preview window showing the center of pressure of
the robot (green) and the support polygon (red).
creating physical replicas of animated,
deformable virtual characters [5]. We
take as input a deformable character
and a set of target poses for it, as illus-
trated in Figure 4. Our method com-
bines finite-element analysis, sparse
regularization, and constrained opti-
mization in order to compute a mini-
mal set of locations where external
forces need to be applied. Further-
more, our method also determines
which regions of the final design need
to be soft, and which ones need to be
rigid, as illustrated in Figure 4. This
inhomogeneous distribution of ma-
terial properties controls the way in
which the physical character will de-
form as external forces are applied to
it. The design output by our framework
can be used as a blueprint to fabricate
physical characters whose range of de-
formations closely matches the target
shapes that were provided as input.
DESIGNING 3-D PRINTABLE
ROBOTIC CREATURES
While the types of functional objects
described thus far are capable of creat-
ing compelling movements, their abil-
ity to interact with the physical world
is limited. In our most recent work,
we bypassed this technical barrier
through an end-to-end solution for cre-
ating 3-D printable robotic creatures
that are able to walk using a variety of
gaits and motion styles [6]. The solu-
tion we developed consists of an inter-
active computational design tool that
allows the user to author both a robot’s
morphology and motion character-
istics. As the user specifies desirable
structural and motion style features,
our system automatically determines
how the robot needs to move such that
it maintains balance at all times.
Our design interface is structured
into two viewports (see Figure 5); one
for editing the structure and motion of
the robot, the other for displaying the
resulting behavior as predicted by our
optimization-based motion model and
through physics simulation. The heart
of the interface consists of a set of easy-
to-use editing tools. The user starts by
loading a description file that specifies
an initial skeletal structure of the ro-
bot, where virtual motors are placed at
each joint position. The user can freely
edit the robot’s structure at all times by
adding or removing motors, thus alter-
ing the morphology of the design, or by
adjusting the position or orientation of
the motors.
The motion of a robotic creature is
completely described by the trajectories

36 XRDS • SPRING 2016 • VOL.22 • NO.3

of its joints. However, authoring mo-
tions directly in this high-dimensional
space is unintuitive, tedious, and very
unlikely to lead to stable movements.
We therefore introduced a set of high-
er-level motion authoring and editing
tools, designed to be mutually orthogo-
nal and intuitive for the user.
Motions are largely characterized
by their footfall pattern, indicating
for each instant during a gait cycle
which of the legs are in contact with
the ground (stance mode) and which
ones are in flight (swing mode). Our
interface displays this information
as a time-dependent graph, allow-
ing for quick inspection and direct
editing by the user. In particular, the
user can change the duration of the
stance and swing phases for any leg
and change the relative ordering of
the footfalls. Not all patterns lead to
desirable motions; this is immediate-
ly clear upon inspecting the evolution
of the support polygon through time
or by observing the motions generat-
ed by our framework. The immediate
feedback provided by our framework
allows the user to interactively adjust
the footfall pattern in order to find
satisfying solutions.
Higher-level goals such as the walk-
ing direction, speed, or turning rate
can be provided by the user so as to
specify the behavior of the robotic crea-
tures that are being designed. Further,
the user can control the overall motion
style by editing the movement of the
robot’s center of mass and feet trajec-
tories. The motions generated by our
optimization-based framework are
guided by this user input, while a set of
feasibility constraints ensures they are
always stable.
Given the structure and motion
goals for a robotic creature, our system
computes time-varying motor values
for dynamically-stable motions using a
trajectory optimization approach. The
user can preview the optimized mo-
tions using physics-based simulation
and iteratively adapt the design to ex-
plore the solution space and converge
on a desired result. Once the design
process is finished, fabrication blue-
prints are generated automatically
such that the robot designs can be
manufactured using 3D-printing and
off-the-shelf servo motors.
XRDS • SPRING 2016 • VOL.22 • NO.3
CONCLUSIONS AND OUTLOOK
In the not-so-distant future, a rich eco-
system of robotic devices will be tightly
integrated into our daily lives. Robots
that help with housekeeping, gardening,
and do-it-yourself projects; assistance
Distinguished
devices for people with disabilities; ro- Speakers Program
bot companions for exploration and
search-and-rescue operations; interac-
tive kinetic art; smart furniture that http://dsp.acm.org
adjusts to individual needs; and robotic
pets for therapy, education, and enter-
tainment—all of these will fundamen-
tally change the way we work, learn, and
play. Many of these robotic devices will
have to be specifically created for differ-
ent tasks, or according to the needs and
preferences of their users. Consequent-
ly, current approaches based on mass- Students and faculty
producing a small number of designs
will become too limited. Therefore, in can take advantage of
conjunction with the continued evolu- ACM’s Distinguished
tion of 3-D printing technologies, com-
putational design tools similar in spirit Speakers Program
to the ones described here have the po-
tential to play an increasingly important to invite renowned
role in how future generations of per-
sonal robotic devices are brought to life. thought leaders in
academia, industry
References
[1] Hecker, C., Raabe, B., Enslow, R., Deweese, J.,
Maynard, J., and Van Prooijen, K., 2008. Real-time
and government
motion retargeting to highly varied user-created
morphologies. ACM Trans. Graph. 27, 4 (2008). to deliver compelling
[2] Coros, S., Thomaszewski, B., Noris, G., Sueda, S.,
Forberg, M., Sumner, R. W., Matusik, W., and Bickel,
B. Computational design of mechanical characters.
and insightful talks
ACM Trans. Graph. 32, 4 (2013).
[3] King, H. C. Planar Linkages and Algebraic Sets. Tech.
on the most important
Rep. math.AG/9807023, 1998.
[4] Coros, S., Martin, S., Thomaszewski, B., Schumacher, topics in computing
C., Sumner, R., and Gross, M. Deformable objects
alive! ACM Trans. Graph. 31, 4 (2012).
and IT today.
[5] Skouras, M., Thomaszewski, B., Coros, S., Bickel,
B., and Gross, M. Computational design of actuated
deformable characters. ACM Trans. Graph. 32, 4 ACM covers the cost
(2013).
[6] Megaro, V., Thomaszewski, B., Nitti, M., Hilliges, of transportation
O., Gross, M. and Coros, S. Interactive design of
3D-Printable Robotic Creatures. ACM Trans. Graph.
34, 5 (2015).
for the speaker
to travel to your event.
Biography
Stelian Coros is an assistant professor in the Robotics
Institute at Carnegie Mellon University. He received his
Ph.D. from the University of British Columbia in 2011. His
doctoral thesis won the Alain Fournier Ph.D. Dissertation
Annual Award. Prior to joining CMU’s faculty, he was a
research scientist working for Disney Research Zurich.
His research lies at the intersection of computer graphics,
robotics, biomechanics, and digital fabrication. More
specifically, Coros is interested in developing models of
motor control, motion planning algorithms, physics-based
simulation tools, and computational design methods for
3-D printable devices. For his work in this area, Coros was
the recipient of a 2015 Intel Early Career Faculty Award.
© 2016 ACM 1528-4972/16/03 $15.00
37
feature

Building a
Toolkit for Fabricating
Interactive Objects
Despite the recent proliferation of easy-to-use personal fabrication
devices, designing custom objects that are useful remains
challenging. RFID technology can allow designers to easily embed
rich and robust interaction in custom creations at low cost.

By Andrew Spielberg, Alanson Sample, Scott E. Hudson,

Jennifer Mankoff, and James McCann
DOI: 10.1145/2889427

F
or years, industry analysts have been predicting the breakout of in-home consumer
fabrication devices such as laser cutters, desktop mills, and most prominently, 3-D
printers. Despite falling costs and bold predictions of their increasing ubiquity,
consumer demand for these devices remains relatively low. Behind this low demand
is the fact that rapid fabrication devices alone are not replacements for mechanical design
expertise or electronics knowledge, making it difficult for laymen to design objects that are
interactive. This limits the typical design space to static objects or simple machines, in turn
limiting the usefulness of these fabrication devices.
Upcoming electronic 3-D printers, doesn’t excessively constrain the design can easily make almost any object in-
such as the Voxel8, coupled with boom- process; and any end user should be teractive with little user expertise. As
ing online maker file sharing commu- able to incorporate it into their designs an added bonus, RFID tags are pow-
nities (such as Thingiverse and 123D regardless of fabrication method. Most ered wirelessly, meaning designers
Make) offer one possible solution to importantly, the solution’s interaction have no messy circuitry to deal with
this problem, potentially putting entire mode needs to be fast in order to guar- and no batteries are required.
suites of electromechanical capabilities antee users true interactivity.
a mere download away. While down- That solution might just be a new AN EXAMPLE APPLICATION
loading designs may be easy, modifying spin on an old classic; a sensor that In order to motivate the types of inter-
them would still likely require the same has long found limited use in the con- active objects we want designers to be
mechanical and electronics expertise sumer transaction and commercial able to make, consider the following
needed to design them in the first place. warehousing industries, but is still scenario. A designer wants to fabricate
To truly empower makers around the a relatively young player in the space a physical Tic-Tac-Toe game board with
world, a better solution is needed. of making and interaction: radio fre- X and O game tokens that are interac-
The solution must be affordable quency identification (RFID). By incor- tive (see Figure 1). When a player places
for hobbyists; it cannot require exces- porating RFID tags in the design proc- a token on the board, a computer run-
sive expert knowledge to design with; ess and managing interactions with ning a companion application provides
it must take up little extra space so it these tags intelligently in software, we auditory feedback. When a player picks

38 XRDS • SPRING 2016 • VOL.22 • NO.3

 up a token, the system detects whether
the token is an X or an O, and the appli-
cation scolds the corresponding player
to wait if it’s not his or her turn. The ap-
plication also keeps track of where X
and O tokens are at all times, displays
the current game with shiny graphics,
and announces when a player has won
or the game has ended in a tie.
There are two parts of this artifact
that must be designed: the method by
which a player interacts with the Tic-
Tac-Toe board, and the specific Tic-
Tac-Toe elements. The ideal solution
should automate everything needed to
implement the game’s interactions—
from geometric design of tokens
through code needed to recognize to-
ken placement and motion—freeing
up the designer to focus on the game
mechanics and aesthetics. As you’ll
see, RFID tags can be the secret sauce
for making this possible.
Image by Andrey Nikolaev
RFID
While you may not necessarily be
aware of how RFID technology works
from a technical perspective, you’ve
almost undoubtedly experienced it at
XRDS • SPRING 2016 • VOL.22 • NO.3
some point in your life. If you’ve ever
seen or used a “tap to pay” credit or
transit card, those interactions rely on
embedded RFID tags to process the
transaction. Large collections of physi-
cal media (such as libraries) use RFID
tags for tracking inventories. Behind
the scenes, RFID tags are used to track
important parcels through storage and
manufacturing processes.
Passive, ultra-high frequency RFID
tags consist of millimeter scale com-
puter chips attached to a (centimeter
scale) antenna, typically embedded in
an adhesive piece of paper. This chip
holds data—a unique identifier and po-
tentially kilobytes of storage. Radio sig-
nals transmitted by an antenna power
tags wirelessly with transmit distances
of up to 10 meters. When a tag receives
a request for information, it reflects a
signal to a reader with the appropriate
response. (See Figure 2 for the compo-
nents). This can be, for instance, its ID
or any of its stored data. A single reader
can be used to track populations of
hundreds of tags. Tags are cheap, cost-
ing less than a 25 cents apiece.
Unsurprisingly, RFID technology
has enjoyed plenty of breakthroughs
in application to ubiquitous sensing;
after all, one of the technology’s killer
apps, transaction processing, is es-
sentially high proximity tag detection.
But other prior work has demonstrated
RFID tags’ potential for tracking hu-
man interaction [1], measuring ges-
tures such as tag touches, swipes on
tag surfaces, tag motion, and even tag
localization. Adding RFID tag-based
interaction can potentially add great
breadth of interactivity to the artifacts
that users fabricate. Further, RFID tags
and their rich vocabulary of interaction
modes can be directly translated into
default widgets to be included into de-
signs, simplifying the design process.
While techniques for measuring
RFID-based interaction are becoming
capable of robustly identifying larger
vocabularies of interaction, increased
robustness has come with a price.
Reliably identifying interactions can
take seconds, which can be prohibi-
tively slow for many real-world interac-
tions. If you were trying to design, say,
a video game controller that uses RFID
tag-based interaction for input, the
39
feature
Figure 1: The target Tic-Tac-Toe game.
difference between a two millisecond
and two-second input lag could be the
difference between player success and
player failure (and frustration). On the
other hand, methods exist with faster
response times, but these methods
are historically less accurate. Misinter-
preting one input for another could be
similarly frustrating and damaging to
usability. Thus, an important design
trade-off arises for developers using
RFID tags: Do you want your applica-
tion to be fast or do you want it to be ac-
curate?
It turns out there’s a way to get the
best of both worlds.
A SOLUTION IN INTERFACE DESIGN
We suspected from previous literature
and experiments that if we wanted in-
teractions to be truly low latency to the
point where they felt natural (say, less
than 200 millisecond latency), intelli-
gently managing this speed-accuracy
trade-off would be at the heart of our
solution. Interaction researchers and
developers have developed a number
of probabilistic methods for managing
uncertainty for more traditional human-
computer interaction (HCI) domains,
including keyboard, mouse, and touch
screen interactions. Could we apply
these methods to our problem of de-
signing RFID powered objects providing
real-time interactions? Previous work
on managing uncertain interactions by
Schwarz et al. [2] represented all inter-
40
Figure 2: An antenna (left) emits radio waves to a 5.3 cm2 RFID tag (right),
which reflects a response to the reader (center).
actions as probabilistic events, and the
applications built atop them were trans-
formed into probability distributions
over program states. For example, if, on
your phone’s touch screen keyboard, you
touched halfway between the “F” and
“G” keys, then their system might repre-
sent the word you were typing as starting
with “F” with a 50 percent probability,
and starting with “G” with a 50 percent
probability. Because of this uncertainty,
their system won’t lock in that first let-
ter just yet; instead it waits for more in-
formation. Subsequent letters then shift
the confidence of the word starting with
“F” or “G” appropriately. If the second in-
put letter is an “H” with 99 percent prob-
ability, then the first letter was probably
a “G.” (No words in the English language
start with “fh,” while many—includ-
ing “ghost,” ghastly,” and “ghoul”–start
with “gh.”) In other words, their system
automatically defers the occasional de-
cisions it’s not confident about, while
maintaining very low latencies for the
typical decisions for which it’s confident.
While a probabilistic representation
of interactions is powerful, exposing it
directly to an application developer is
dangerous. The average person is bad
at probability. Asking a user to calcu-
late explicit probabilities of higher-
level states (such as the probability
of certain sentences in our phone ex-
ample) would never gain traction in an
API. So, the clever idea Schwarz et al.
had was to have developers specify de-
terministic interactions, keeping the
programming of states intuitive, and
let the API automatically manage the
probabilities behind the scenes. Their
program state distribution is repre-
sented as a collection of potential state
samples, and those state samples (and
thus the distribution) are updated with
input samples drawn from input events
such as key presses. Those input events
are also represented probabilistically.
This sampling approach is also known
as Sequential Monte Carlo sampling
(SMC) or particle filtering.
SMC is not specific to any single type
of input; it can even manage inputs
from several input modalities at once.
So naturally, the SMC framework could
extend to our RFID-based scenario if we
considered RFID tag reads (or the lack
thereof) as individual inputs (similar
to key presses in the phone example).
Ideally, such a system could detect pro-
gram states it were unconfident about,
then it could defer making decisions
about them—perhaps only by milli-
seconds—and avoid misclassification.
SMC could help measure and manage
these confidences. However, applying
the SMC method would mean we would
need to model interactive input with
RFID tags probabilistically. What types
of inputs would we want to detect, and
where would their models come from?
TOUCH AND GO
We decided early on to focus on mod-
XRDS • SPRING 2016 • VOL.22 • NO.3
eling two modes of interaction for fab- model, and p(z1:i) is a normalization stop occurring, and so the presence
ricating interactive devices: 1) touch factor, constant over all hypotheses, of a touch event is actually described
events, that is, when a user physically which can typically be ignored. In our by the absence of data. When a tag is
touches and covers a tag; and 2) mo- setting, the measurement model has covered, the reader goes from to read-
tion, that is, the velocity with which a far bigger influence over the state ing its presence every 50-200 millisec-
a tag is moving relative to a reader. estimate than the prior, and many onds or so to every three seconds at
These inputs would allow us to build priors work fine in practice. However, best. Thus, as more time passes be-
a design API that would allow for we found an intuitive solution is to tween subsequent tag reads, the less
reasonably large variety in interac- simply bias the state estimate toward likely the tag is actually visible.
tive objects, including touch menus, increasing uncertainty. We performed a number of experi-
token-based games, spinners, slid- Now, you may be thinking that ments over various tag population siz-
ers, accelerometers, and so on. For all of this is overkill. After all, RFID- es where we recorded times between
example, in our Tic-Tac-Toe game, based credit card transactions also subsequent tag reads, including sce-
these tag interactions alone could be rely on whether or not a tag is visible narios where the tags were covered
used to implement both token place- to detect interaction, and they trigger and some scenarios where they were
ment identification (using tag touch/ almost instantaneously as soon as a uncovered. From the data, we were
cover) and token motion measure- reader sees the tag. But that instan- able to build probability distributions
ment (using tag motion). Further, in- taneous transaction relies on a tag of the times between tag reads in each
put and touch events lend themselves transitioning from a default state of of the covered and uncovered states,
well to salient features measured by invisible to a reader, to visible. Mean- providing our measurement model.
readers about tags. For touch, the while, touch events begin when a tag Coupled with our Bayesian filter and
time between each tag’s consecutive is suddenly occluded, transitioning a prior, which rebiases the state es-
reads provides a strong indicator as from a default state of visible to invis- timate toward a 50–50 estimate of a
to whether or not the tag is occluded ible. When touched, reads for a tag covered/uncovered tag state, we can
by a conductive material such as foil
or a dielectric material such as skin. Figure 3. RapID’s data-to-application pipeline.
For motion, the rate of change of the
phase of the received radio wave for
each tag gives strong clues as to how
fast any given tag is moving.
Consecutive inputs are correlated.
If a tag is moving with a certain veloc-
ity during one tag read, it will likely be
moving with a similar velocity at the
next read. Since tag touch events are
far less frequent than tag state reads, it
is likely a covered tag will remain cov-
ered between reads (and likewise for
an uncovered tag). Therefore, we de-
cided to keep running measurements Figure 4: Our Sketchup design environment; creating Tic-Tac-Toe.
of tag states using Bayesian filters. A
Bayesian filter fuses previous state esti-
mates with new measurements to con-
stantly provide robust measurements of
noisy systems. It’s called a Bayesian filter
because it does this through recursive ap-
plications of Bayes’ rule. Given a sequence
of i – 1, state estimates x1:i–1 ={x1,x2…xi–1},
and sequence of i observations
z1:i={z1,z2 ,…zi} a Bayesian filter esti-
mates the ith state as

p(zi|xi)p(xi|xi-1)
p(xi|z1:i, xi-1) =
p(z1:i)

Here p(xi|xi-1) is a hand-tuned Bayes-

ian prior, which represents state evo-
lution in the absence of observation,
p(zi|xi) is known as the measurement

XRDS • SPRING 2016 • VOL.22 • NO.3 41

feature
keep continuous estimates of tag state
in the absence of tag reads, as well as
update our distribution with more cer-
tainty when a tag read arrives. Using
this method, we are able to measure
up to 20 tags at once with latencies of at
worst 200 milliseconds (and typically,
our state estimation converges much
faster than that).
Measuring motion is a bit different.
Motion can only be measured when a
tag is visible, so in this domain a reader
is constantly receiving reads. There-
fore, accurately measuring motion
depends on processing information
about the tag read, not just if and when
a tag read happens.
In order to measure motion, we
used some results in physics that state
the velocity v of a tag is proportional to
the change in phase Δφ between reads
divided by time between reads Δt (nor-
malized by the frequency f of the car-
rier wave). In other words:
Δφ
v= ∝
f*Δt
This was a continuous domain
rather than the discrete binary visible
versus not visible domain of the touch
event, so we had to record tag motion
over many different velocity measure-
ments. From this, we were able to
build an empirical Gaussian of mea-
sured phase changes given known ve-
locities, which, coupled with the run-
ning Bayesian filter, provides robust
velocity measurements in practice
(with built in error bars). In this case,
we use a prior that maintains the pre-
vious velocity estimate and increases
its variance, indicating increasing un-
certainty in the estimate without in-
Figure 5. Our Pong demo application and sliders.
42
formation. Similar to the touch event
scenario, updates are fast, and we are
able to measure up to 20 tag velocities
at once with latencies of at worst 200
milliseconds.
AN API FOR FABRICATION
With the technical details of how we
could formulate RFID interactions as
probabilistic events out of the way, we
were able to carefully consider what
our API (which we are calling “Rap-
ID”) for fabricating interactive objects
should look like. Figure 3 shows the en-
tire pipeline of how data is translated
into interaction with an application.
But how should developers create expe-
riences in such a paradigm? We knew
we wanted developers to be able to cre-
ate interactive programs, using RFID
tags as inputs. We also knew the API
for developing these programs should
abstract the notion of probabilistic
program states away from the users.
Finally, in the same way we abstracted
away much of the electromechanical
design using RFID tags, we would need
to abstract away as much coding as
possible for novice programmers.
The powerful decision is to couple
functionality with pre-defined physi-
cal widgets, combining the fabrication
and interactive experience directly.
This leads to two levels of fidelity with
which developers could design physi-
cal user experiences.
At the high level, we developed a
physical design environment as an ex-
tension to SketchUp, which is a CAD
environment aimed at novice users.
Figure 4 shows a user designing Tic-
Tac-Toe with our design environment.
Figure 6. Our spaceship demo application.
We designed a number of physical wid-
gets users can add in order to make
their designs interactive: tokens, spin-
ners, sliders, touchable surfaces, veloc-
ity sensors, and so on. After users final-
ize their design, they can export their
designs to digital files for fabrication
via 3-D printing or laser cutting, as per
usual. RapID also exports two other
important pieces of information.  
First, in order to ease the program-
ming of the digital side user applica-
tions, RapID exports starter code,
which can be immediately run. This
starter code keeps running estimates
of widgets—which tokens are placed
in which slots, the position of sliders
and spinners, how fast an object is
moving, and so on—and registers call-
back functions to monitor changes in
these widgets’ states. All users need
to do is define these callbacks deter-
ministically to say how the program
state should be updated when the tag
states change. RapID then uses those
deterministic functions to update
the probabilistic program state us-
ing SMC. RapID widgets also provide
visualization methods for providing
on-screen visual feedback of the prob-
abilistic state deterministically. For
example, RapID can render objects us-
ing the mean state for velocity-based
widgets, or for tokens, render based
on the most probable placement con-
figuration. Our API is built with Unity,
making it easy to build beautiful in-
teractive media built on top of it. Sec-
ond, in order to ease fabrication and
assembly, RapID annotates design
files with the locations where users
should place RFID tags, along with
XRDS • SPRING 2016 • VOL.22 • NO.3
the IDs with which those tags should
be programmed.
At the low-level, though, we recog-
nized our pre-defined widgets may not
be expressive enough for all applica-
tions. That’s why, for the experienced
users, we exposed the lower-level API
for interacting with the probabilistic
program state. In order to make it pos-
sible for experienced users to develop
their own physical widgets and their
associated code.
While this project is by far not the
first to allow users to build physical wid-
gets that digital programs can be built
on top of [3, 4], the fact that RFID tags,
which are small and thin, have very few
geometric constraints makes it very
easy to place them anywhere in designs.
This makes it easy to grow large, expres-
sive widget libraries. In the future, it
will be exciting to see how RFID tags
and other similar, versatile sensors, will
allow online communities to grow large
widget libraries much in the way maker
communities such as Thingiverse cur-
rently share pure .STL files.
PUTTING IT ALL TOGETHER
For now, we’ve created a few demo ap-
plications to show off the promise of a
toolkit, which is a synthesis of our ap-
plication pipeline (see Figure 3) and
Sketchup Front-End (see Figure 4).
Using RFID tags on tokens and to-
ken slots, we were able to build a wire-
less, low-latency, physical game of
Tic-Tac-Toe. Here, we used our token
widget, which places tokens opposite
conductive foil to measure whether
or not token/slot pairs are visible. Us-
ing the IDs of the tags, we can iden-
tify which token is placed, when it is
placed, and where it is placed. When
the widget is added to the design, our
Sketchup extension adds the appropri-
ate token and slot geometry to the digi-
tal design files, and automatically gen-
erates all of the code for tracking this
interaction. The only code the user has
to add is the traditional deterministic
game of Tic-Tac-Toe, and the visual
and auditory feedback for the players,
all of which can be written in fewer
than 100 lines of C# code.
In another example (see Figure 5),
we used our slider widget, which fea-
tures a conductive cover that slides
atop a line of RFID tags. Our automati-
XRDS • SPRING 2016 • VOL.22 • NO.3
cally generated interaction code esti-
mates the state of the slider based on
which tags are visible to the RFID read-
er and which are masked by the cover.
We 3-D printed one controller and laser
cut the other (just to show we could),
and painlessly coded up a flashy demo
of the classic arcade game Pong using
our sliders as wireless controllers.
As a final example (see Figure 6),
we demonstrated our RFID tags’ mo-
tion sensing capabilities with a simple
spaceship-based demo. We designed
a simple toy spaceship and placed a
raw tag widget on the design, which,
while not adding new geometry to the
design, generated code for touch and
motion callbacks. This demo is partic-
ularly friendly to novice programmers.
Using our API, it was easy to translate
our toy’s motion to the digital on-
screen motion of a virtual spaceship,
only writing new code for on-screen
animation. (Dong Nguyen, we eagerly
anticipate your Flappy Bird port for our
RFID-based system!)
CONCLUSIONS
As RFID tags become more robust and
tag readers become cheaper with each
passing year, RFID sensing is rapidly be-
coming a serious contender for making
physical fabrication projects interac-
tive. RFID sensing provides a platform
that is easy to design with and even
easier to interact with and use. A future
where anybody can quickly fabricate
wirelessly powered novel game control-
lers, smart-home devices, personal ro-
bots, and more is right around the cor-
ner. It will be exciting to see how other
sensors can be hacked through simi-
lar data-driven methods, to go beyond
their original intended purpose for use
in interactive fabrication projects.
Through a combination of inexpen-
sive, easy-to-use sensors, and more sys-
tems that marry physical design with
digital design, people will finally feel
empowered to make devices based on
how they are meant to be used, and
not just on how they are meant to look.
Novice makers will finally be able to
design and fabricate devices that fully
capture the interactive nature of their
imagination. And when interactive ob-
jects are as easy to make as static ones,
the personal fabrication movement
will truly be ready to take off.
ACKNOWLEDGMENTS
This work was conducted as part of a re-
search project at Disney Research: Pitts-
burgh. This work is based on a paper to
appear in the CHI 2016 proceedings.
References
[1] Li, H., Ye, C., and Sample, A. P. IDSense: A human
object interaction detection system based on
passive UHF RFID. In Proceedings of the 33rd Annual
ACM Conference on Human Factors in Computing
Systems (CHI 2015) (April 18–23, Seoul). ACM, New
York, 2015, 2555–2564.
[2] Schwarz, J., Mankoff, J., and Hudson, S. E. Monte
Carlo Methods For managing interactive state,
action, and feedback under uncertainty. In
Proceedings of the 24th Annual ACM Symposium on
User Interface Software and Technology (Oct. 16-19,
Santa Barbara, CA). ACM, New York, 2011, 235–244.
[3] Greenberg, S. and Fitchett, C. Phidgets: Easy
development of physical interfaces through
physical widgets. In Proceedings of the 14th Annual
ACM Symposium on User Interface Software and
Technology (UIST 2001) (Nov. 1114, Orlando). ACM,
New York, 2001, 235–244.
[4] Laput, G., Brockmeyer, E., Hudson, S. E., and
Harrison, C. Acoustruments: Passive, acoustically-
driven, interactive controls for handheld devices.
In Proceedings of the 33rd Annual ACM Conference
on Human Factors in Computing Systems (CHI
2015) (April 18–23, Seoul). ACM, New York, 2015,
2161–2170.
Bigographies
Andrew Spielberg is a second-year Ph.D. student at the
Computer Science and Artificial Intelligence Laboratory
(CSAIL) at MIT, where he works in the intersection of
fabrication and robotics. His current research exploits
data-driven methods for optimizing the design and behavior
of 3-D printed robots. His prior research has focused on
automated assembly. Prior to joining MIT he received his B.S.
and master’s from Cornell University and spent time at The
Johns Hopkins’ Applied Physics Laboratory.
Alanson Sample is a research scientist at Disney Research,
Pittsburgh where he leads the Wireless Systems group.
His research focuses on enabling new guest experiences
and sensing and computing devices by applying novel
approaches to electromagnetics, RF and analog circuit
design, and embedded systems.
Scott Hudson is a professor of human-computer
interaction in the School of Computer Science at Carnegie
Mellon University, where he serves as the founding
director of the HCII Ph.D. program. He received his Ph.D. in
computer science from the University of Colorado in 1986,
and has previously held faculty positions at the University
of Arizona and the Georgia Institute of Technology. Elected
to the CHI Academy in 2006, he has published extensively
on technology-oriented HCI topics, and recently received
the Allen Newell Award for Research Excellence at CMU.
Jennifer Mankoff is an associate professor in the Human
Computer Interaction Institute at Carnegie Mellon
University. She earned her B.A. at Oberlin College and
her Ph.D. in computer science at the Georgia Institute of
Technology. Her research enhances the human experience
with technology. Her goal is to combine empirical methods
with technological innovation to construct middleware
(tools and processes) that can enable the creation of
impactful applications. Most recently, this work has
focused on 3-D printing and its potential for creating
custom assistive technologies for people with disabilities.
James McCann obtained his Ph.D. in 2010 from Carnegie
Mellon University.  His research hours are spent at Disney
Research Pittsburgh developing systems and interfaces
that operate in real-time and build user intuition; lately,
he has been dabbling in the creation of physical objects.
He also makes video games as TCHOW llc, including recent
releases “Rktcr” and “Rainbow.”
© 2016 Copyright held by Owner(s)/Author(s).
Publication rights licensed to ACM.
1528-4972/16/03 $15.00
43
feature

3-D Printing
Interactive Objects
Today’s 3-D printing hobbyists churn out kilos of static trinkets.
These existing machines can further help them create functional
objects, if new perspectives and designs are employed.

By Valkyrie Savage
DOI: 10.1145/2893495

3
-D printers have evolved from professional equipment for industrial design studios
to staples in makerspaces—these community hubs offer open access to fabrication
machines, electronics tools, and more. As 3-D printers make their way into clubs,
schools, libraries, museums, and homes, we should consider what sorts of objects
can these machines make and how will users design those objects?
A look at Thingiverse, an online 3-D printing community, reveals novices mostly create
static and decorative objects—figurines, ornaments, and pen or tool holders—stuff that
hardly belongs to a new “industrial revolution.” In contrast, experienced designers create
functional objects, often as assemblies. mentation and assembly. Physical tive sensing strategies that only re-
These include existing, non-printed user interfaces are pervasive—think quire adding a single sensor to a 3-D
electronic parts such as sensors and ac- of game controllers and musical in- printed model to recognize a range of
tuators, and mechanical parts such as struments—and their physicality has different interactions.
fasteners and hinges. Designing elec- important benefits such as tactile
tro-mechanical assemblies with func- feedback and high-performance ma- HANDS-ON DESIGN
tional components in CAD remains a nipulation [1]. For example, gamers Toolkits, such as Arduino or .NET
complicated task for experts. Design- prefer physical input for speed and Gadgeteer, have lowered the threshold
ers must specify not only where to place performance, while musicians are of experimenting with electronics for
functional components, but how to interested in virtuosity and control. interaction, at the same time creat-
mount them to allow assembly and en- Building working devices that also ing a community and extensive docu-
sure functionality. exhibit interactive behavior requires mentation. This has allowed artists,
At Berkeley Institute of Design adding electronic sensing compo- students, and other non-professionals
(BiD Lab), our research goal is to help nents and circuitry to the mechanical to leverage capabilities such as lights,
craft electronics for hobbyists of all design. We are exploring two direc- sound, and sensing in their projects
Image by Vitaliy Zuyenko

ages, often called “makers,” to cre-
ate functional interactive objects on
commodity digital-fabrication ma-
chines. We seek to enable makers to
print working physical user interfac-
es with minimal additional instru-
tions to make this task easier. First,
we are building tools that automati-
cally fit electronic components into
3-D printed models. Second, we are
investigating ways to replace stan-
dard wired electronics with alterna-
[2]. However, integrating electronics
into 3-D printed objects is difficult.
Mounting parts, such as buttons and
joysticks, in exactly the right place
may require significant changes to a
3-D model (e.g., to add fasteners and

44 XRDS • SPRING 2016 • VOL.22 • NO.3

Photo Credit TK

XRDS • SPRING 2016 • VOL.22 • NO.3 45

feature

Figure 1. With the Makers’ Marks system, makers create and annotate physical designs of objects (left). Once these are 3-D
scanned, the annotations are replaced with relevant geometry (center), and can be printed for components to snap in (right).

clearances, or to split an enclosure cision of CAD. be accessible to the end-user. By em-

into two half shells).
We ran a formative study that sug-
gests novices can express their intent
for the design of physical interfaces by
combining physical sculpting of larger
shapes and annotation of finer details,
for instance, through additional draw-
ings or marks. This study gave us the
inspiration to consider tangible mod-
eling as an alternative to CAD.
Inexpensive 3-D scanning tech-
nologies are becoming increasingly
available based on smartphones [3]
or web cams. 3-D scanning combined
with 3-D printing opens up the pos-
sibility of a new workflow. A maker
may start with a modeling task in the
physical space, followed by scanning
the resulting object to get a digital rep-
resentation. Then that digital model is
modified algorithmically, and finally
printed back out. This physical–to-
digital-to-physical pipeline could com-
bine some of the benefits of tangible
modeling with the flexibility and pre-
We created a system, called “Mak-
ers’ Marks,” which allows users to
physically design custom enclosures
with precision mounting geometry
for functional components. To design
the overall shape, users first sculpt
using clay or other physical materi-
als. Then, they add annotations with
physical stickers indicating place-
ment for functional components. In
our prototype, supported parts in-
clude components for user interaction
(e.g., joysticks) as well as mechanical
parts (e.g., hinges). These stickers are
designed with two purposes in mind:
They should be legible to people, as
well as reliably detectable using com-
puter vision algorithms.
Makers’ Marks captures user-creat-
ed geometry using a 3-D scanner and
replaces annotations with precise 3-D
geometry from a library. The gener-
ated geometry ensures the indicated
components can be fastened in by the
designer post-print, and, hence, will
ploying physical authoring for rough
shapes and digital tools for precise as-
sembly geometry, we hope to enable
easier, quicker creation of complex
functional objects (e.g. a game con-
troller, shark, box, and baby monitor),
as shown in Figure 1.
Makers’ Marks creates rigid,
shelled objects that incorporate
surface-mounted functional compo-
nents. These components must have
3-D models and additional clearance
metadata. In spite of these caveats,
objects created with this technique
can range from whimsical and deco-
rative (a waving shark) to precise and
functional (a game controller). The
physical-to-digital-to-physical pipe-
line enables a wide variety of device
designs, which is limited only by the
maker’s imagination.
Other projects have also offered
ways to integrate existing parts into
novel 3-D models. For example, the
faBrickation technique accelerates

Figure 2. Sauron’s processing steps start with a hollow 3-D model (a). When a virtual camera model is added, it can detect
which components are visible and not visible (b), and modify invisible components as necessary (c). Once the object is printed
(d), a single camera can view (e) and track (f) the user-facing components.

46 XRDS • SPRING 2016 • VOL.22 • NO.3

3-D printing processes by 3-D print-
ing high-detail areas around assem-
bled low-resolution LEGO blocks [4].
The Enclosed interface allows users
to interactively design enclosures
around pre-existing electronics [5].
With Makers’ Marks and other
related projects, a user assembles
a number of different off-the-shelf
electronic and mechanical parts into
a 3-D printed shell. Is it possible to
further reduce this complexity by le-
veraging the flexibility of 3-D print-
ing? Could it be possible to instan-
taneously transform a 3-D print into
an interactive device by adding only a
single “super sensor”?
GENERATING 3-D GEOMETRY FOR
INTERACTIVE SENSING
3-D printed plastic has a number of
properties that can be leveraged for
designing useful sensing techniques.
Through the printing process, it
can be laid down in arbitrary geom-
etries—mechanisms can be printed
fully functional without needing as-
sembly—and multiple colors or mate-
rials can be used. We have developed
tools and strategies to utilize several
of these properties. Sauron is a de-
sign tool that enables users to rapidly
turn 3-D models of input devices into
interactive 3-D printed prototypes by
adding a camera pointed into the in-
side of a device. Our system automati-
cally modifies the shape and color of
mechanisms so interaction can be
sensed with standard computer vi-
sion techniques. Lamello is a tech-
nique that senses interaction with a
single microphone. It generates 3-D
printed tine structures, which are
struck by the user and then vibrate at
predictable frequencies.
Sauron: The all-seeing camera. Sau-
ron enables makers to 3-D print a com-
plete interactive device in a single step.
After printing, they add a miniature
camera with an integrated ring light
to the prototype. After an interactive
registration step, Sauron can track the
motion and position of buttons, slid-
ers, joysticks, and other input devices
through machine vision performed on
an end-user’s computer.
Sensing all input components on
a device with complex shapes can
be challenging, as components may
XRDS • SPRING 2016 • VOL.22 • NO.3
be outside the viewing frustum of
a single camera or blocked by the
device’s geometry. To address such
challenges, we introduce automatic
visibility analysis and model modi-
fication to translate human input
into visible movement, which can
be accurately tracked with standard
computer vision algorithms. Our
system first determines which com-
ponents will be visible to the camera
based on a maker’s placement of a
virtual camera into the CAD model
during the design phase (see Figure
2b). For components that are not
visible, Sauron can modify the com-
ponent model’s internal geometry
to extend motion into the camera’s
viewing frustum using parameter-
ized extrusions (see Figure 2c). Next,
Sauron uses ray tracing to determine
how optical mirrors may be placed to
make motion visible in cases where
geometry modification fails because
of mechanical interference. We im-
plement these techniques by extend-
ing commercial parametric CAD
software. The models can be printed
on any 3-D printer that has remov-
able support material (see Figure
2d). While computer vision research
traditionally strives to uncover infor-
mation about an unknown environ-
ment, our approach seeks to modify
a known environment. In this case,
the digital model of the prototype
object to be fabricated in order to fa-
cilitate computer vision (see Figures
2e and 2f).
To evaluate Sauron’s expressiv-
Figure 3. Components sensed by Lamello have tines printed at interaction points
(e.g., under the stroking path of a slider), which vibrate at predictable frequencies.
ity, we created a series of functional
prototypes—a game controller, ergo-
nomic mouse, and DJ mixing board.
We also asked three knowledge-
able CAD users to design DJ mixing
boards with our sensing approach in
mind. In all cases the users were able
to focus on the usability of their pro-
totype interfaces without being im-
peded by the sensing technique. Ad-
ditionally, we evaluated 10 pre-made
models downloaded from the Inter-
net. We determined even designers
who did not have vision-based sens-
ing in mind while designing would
have been able to use Sauron for their
prototypes in most cases.
The Sauron approach was inspired
by prior research on 3-D printing light
pipes and integrating optical sensors
into prints [6]. It has some important
assumptions and limitations: Our im-
plementation of the CAD plugin can
currently only process certain types
of hollow models and is not guaran-
teed to succeed. Second, our current
model modification techniques only
work for a subset of input compo-
nents, though they are extensible.
Despite these limitations, Sauron en-
ables construction of a useful variety
of devices. However we wanted to ex-
plore additional sensing techniques,
which could leverage other plastic
properties and different sensors.
Lamello: A tine-y technique.
Lamello integrates algorithmically-
generated tine structures into mov-
able components to create passive
tangible inputs like the slider, button,
47
feature
and dial (see Figure 3). Manipulat-
ing these inputs creates sounds that
can be captured using an inexpensive
contact microphone or existing laptop
microphone, and then interpreting
those sounds using real-time audio
signal processing. Lamello predicts
the fundamental frequency of each
tine based only on its digital geom-
etry. Thus, recognition does not re-
quire training examples. The decoded
high-level events can then be forward-
ed to interactive applications. The
name “Lamello” is derived from the
lamellophone family of instruments,
which create sound through vibrating
tongues of varying lengths.
There are two main challenges in
developing a passive acoustic sens-
ing technique that supports a vari-
ety of input controls: designing the
physical mechanisms for generating
sounds, and developing recognition
algorithms that can interpret those
sounds in the intended manner.
To generate sounds, we embed
tine structures in input components.
Our tines are rectangular beams at-
tached at their base to the compo-
nent, and free to deflect at their top.
When an end user interacting with
a component causes tine plucks, the
tines vibrate the body of the compo-
nent and the vibrations are captured
by the microphone as sound. Tines
can be arranged in configurations
supporting different interactions
(e.g., sliding, rotating, pressing). The
audio signal of a tine strike is char-
acterized by an initial transient—a
short, high-energy sound across a
wide range of frequencies followed
by free vibration with a local long-de-
cay energy peak at the tine’s resonant
frequency. Conceptually, our recog-
nizer detects a transient, finds the
dominant resonant frequency after
the transient passes, and compares it
to predicted tine frequencies.
Recognizing mechanically gener-
ated sound for input has important
limitations. Movement generates
sound—meaning steady state cannot
be sensed. At the same time it also has
appealing characteristics: compatible
components can be fabricated from a
single material (e.g., 3-D printed ABS
plastic), and “wiring” just requires
attaching a microphone. We also de-
48
We seek to enable
makers to print
working physical
user interfaces
with minimal
additional
instrumentation
and assembly.
veloped specific design and fabrica-
tion guidelines, and demonstrated
several components that use the Lam-
ello approach. Our evaluation showed
training-free recognition is possible,
though our accuracy could be im-
proved with de-noising techniques.
Other researchers are exploring re-
lated techniques that characterize the
sound of surface textures when a user
scratches an object [7]. Laput et al. use
an active sensing approach in which
a speaker continuously plays sound
through a hollow, flute-like pipe that
ends in a microphone. When end-
users manipulate 3-D printed mecha-
nisms along the pipe, they change its
acoustic properties [8].
WHAT’S NEXT?
Our work has explored different
methods to create functional and
interactive objects through tangible
modeling and through novel sensing
techniques. Emerging fabrication
technologies may soon open addi-
tional areas of exploration. We now
have machines that can either spray
conductive material on the outside
of existing objects, or deposit con-
ductive material layer-by-layer inside
3-D printed objects [9, 10]. With such
machines, traditional printed circuit
boards and wiring may become obso-
lete and the object itself may become
the circuit. In addition, advances in
material science, such as continuous
liquid interface printing, promise to
cut down printing time by an order
of magnitude [11]. Cheap and fast
fabrication could make it possible to
explore larger design spaces of inter-
active objects by automatically gener-
ating alternatives and printing them
side-by-side for comparative testing.
All this new hardware will require
new design software. One intriguing
opportunity lies in automatic gener-
ation (or modification) of interfaces
to suit particular people and tasks,
e.g., for assistive applications. We
believe there is significant territory
to be explored in modeling users’
individual capabilities, as well as
understanding how to create opti-
mal inputs devices suited to specific
tasks. Software is playing an increas-
ingly important role in defining the
physical objects that will surround
us in the future. It is an exciting time
to be a computer scientist.
References
[1] Klemmer, S. R., Hartmann, B. B., and Takayama, L.
How bodies matter: Five themes for interaction
design. In Proc. Conference on Designing Interactive
Systems (DIS ’06). ACM, New York, 2006, 140-149.
[2] Arduino. https://www.arduino.cc
[3] Autodesk 123D Catch. http://www.123dapp.com/catch.
[4] Mueller, S., Mohr, T., Guenther, K., Frohnhofen, J.,
and Baudisch, P. faBrickation: Fast 3D printing
of functional objects by integrating construction
kit building blocks. In Proc. Conference on Human
Factors in Computing Systems (CHI ’14). ACM, New
York, 2014, 3827-3834.
[5] Weichel, C., Lau, M., and Gellersen, H. Enclosed:
A component-centric interface for designing
prototype enclosures. In Proc. International
Conference on Tangible, Embedded, and Embodied
Interaction (TEI ’13). ACM, New York, 2013,
215–218.
[6] Willis, K., Brockmeyer, E., Hudson, S., and Poupyrev,
I. Printed optics: 3D printing of embedded optical
elements for interactive devices. In Symposium on
User Interface Software and Technology (UIST’ 12).
ACM, New York, 2012, 589–598.
[7] Murray-Smith, R., Williamson, J., Hughes, S., and
Quaade, T. Stane: Synthesized surfaces for tactile
input. In Proc. Conference on Human Factors in
Computing Systems (CHI ’08). ACM, New York, 2008,
1299–1302.
[8] Laput, G., Brockmeyer, E., Hudson, S. E., and
Harrison, C. Acoustruments: Passive, acoustically-
driven, interactive controls for handheld devices. In
Proc. Conference on Human Factors in Computing
Systems (CHI ‘15). ACM, New York, 2015, 2161–
2170.
[9] Optomec. http://www.optomec.com
[10] Voxel8. http://www.voxel8.co/
[11] Carbon. http://carbon3d.com
Biography
Valkyrie Savage is a Ph.D. candidate at UC Berkeley.
Her research focuses on design tools for 3-D printing,
specifically for creating interactive objects like video game
controllers. She is broadly interested in technologies to
encourage interest and participation in STEAM.
© 2016 Copyright held by Owner(s)/Author(s).
Publication rights licensed to ACM.
1528-4972/16/03 $15.00
XRDS • SPRING 2016 • VOL.22 • NO.3
 INSPIRING MINDS
FOR 200 YEARS
Ada’s Legacy illustrates the depth
and diversity of writers, things, and
makers who have been inspired
by Ada Lovelace, the English
mathematician and writer.
The volume commemorates the
bicentennial of Ada’s birth in
December 1815, celebrating her
many achievements as well as
the impact of her work which
reverberated widely since the late
19th century. This is a unique
contribution to a resurgence in
Lovelace scholarship, thanks to the
expanding influence of women in
science, technology, engineering and
mathematics.

ACM Books is a new series of high quality books for the computer science community, published by
the Association for Computing Machinery with Morgan & Claypool Publishers.
feature

Soft Printing
with Fabric
3-D printed objects made of fabric could be flexible and
deformable, bringing possibilities to new sensors and actuators.

By Huaishu Peng, Scott Hudson,

Jennifer Mankoff, and James McCann
DOI: 10.1145/2893499

A
s more consumer level 3-D printers enter the end-user market, we have seen a boost
in 3-D printing that empowers users to print customized 3-D models on demand.
Glance over to the Thingiverse, one of the biggest online 3-D printing document
sharing platforms, where you can find almost whatever you want: functional objects
like VR Goggle frames or a tourbillon watch, artistic artifacts like a spaceship model, or, of
course, various versions of classically scaled octopus models.
Moving beyond hobbyist commu- ize they are not that great. Most of the Stratasys Objet260). In actuality, 3-D
nities, 3-D printing has also found its artifacts are made of hard, rigid plas- models can be printed very accurately
way into numerous industries, includ- tic. There is nothing wrong with rigid with such material. Nike, Feetz, and a
ing fashion. Last April, the Inside 3D prints, if you want an egg holder or a few other fashion companies have been
Printing Conference and Expo hosted a lightsaber. However, when you spend working to create 3-D print customized
curated 3-D print fashion show in New a couple of hours printing a teddy shoes and insoles with this rubber-
York City that featured various apparel bear,that feels like stone, or when you like material. Another option is ther-
and accessories—from dresses full of realize the fabulous 3-D dress on the moplastic elastomers, recently made
flowing curves to a dreamy bowler hat runway won’t comply to your body and available in the form of filament wires,
with horns. The idea of digitally custom- is uncomfortable to boot, you know which allow consumer-grade 3-D print-
izing one’s clothing and having it di- there is room for improvement. ers to print soft goods as well. Finally, if
rectly printed using a fabricator sounds 3-D printing material libraries are one would like to fiddle with structure
appealing not only to consumers, but missing one important material type; design, you can achieve bending prop-
to designers as well. For example, Ki- material that is soft, deformable, and erties with 3-D printed living hinges.
nematics, designed by Nervous System, gives you the feeling of real cloth. These are just some of the classes of
is a folding mechanism system that en- 3-D printing material that enable new
ables 3-D printing of flexible garments SOFT PRINTING MATERIAL design opportunities for printing soft
that move and sway like real cloth. Admittedly, soft printing materials are objects or to achieve bending effects.
All of this may look great from afar. available in the 3-D printing market. A However, looking at the diversity in
But when you get a little closer and UV-cured silicone material that resem- hard materials, we need more than one
sweep your finger across these care- bles rubber is available for high-end, type of soft material that moves away
fully designed 3-D garments, you real- industrial-level machines (such as the from the look and feel of rubber. Yes we

50 XRDS • SPRING 2016 • VOL.22 • NO.3

 want customized shoes printed from
soft silicone or rubber, but we would
also love to have 3-D toys that are direct-
ly fabricated from a 3-D printer and are
also soft and cuddly.
Toward this end, our contribu-
tion is two new technologies that can
print 3-D geometries with fabric. To
be more specific, we will look at 3-D
printing with the two essential forms
of fabric: yarn and fabric sheets. Since
fabric property is different than most
conventional raw materials used in
3-D printing (usually powder, liquid,
or filament), we have developed our
own 3-D printers to handle fabric. In
the following sections, we detail the
design of two different types of fabric
printers, the challenges of printing
with soft fabric, and our solutions to
address those concerns. Finally, with
the ability to print with fabric mate-
Image by Lyly / shutterstock.com
rial, we move beyond and explore the
potential use of soft print, both me-
chanically and electrically. The goal
of our project is to bring the sensa-
tion of softness and the intimate feel
of fabric to 3-D printing techniques,
eventually creating one-off interac-
tive soft objects.
XRDS • SPRING 2016 • VOL.22 • NO.3
SOFT PRINTING WITH FELTED YARN
The exploration starts with Scott Hud-
son’s work, printing felt in the form of
yarn [1]. As human beings, we have a long
history of making clothes using balls of
yarn. Knitting or weaving yarn into dif-
ferent patterns, either by hand or with
machines, creates three-dimensional
garments that are flexible to fit our body.
However, Hudson’s 3-D printing tech-
nique is to fabricate 3-D solid geometries
out of soft yarn. Due to the soft property
of felted yarn, the end results are not the
expected precise shapes associated with
normal hard-printed objects. Instead,
his project emphasizes 3-D objects with
not only varied textures, but with the feel
of hand-crafted fiber arts reminiscent of
felted yarn material.
3-D printing with felted yarn creates
challenges, such as: How to form 3-D
geometries with such flexible material,
and how to bond layers? Conceptually,
the 3-D felt printer design is very simi-
lar to a Cartesian 3-D printer; there is a
three-dimensional motion platform as
part of the printer frame, and a custom-
ized printing head handles the printing
material. To print a 3-D geometry, the
printer software slices the model in a lay-
ered fashion, and the printer lays down
a series of thin layers of fabric each rep-
resenting a horizontal image of the fi-
nal geometry. By building the geometry
from the ground up and bonding each of
the 2-D layers, a complete 3-D object with
fairly arbitrary geometry can be formed.
Due to the similarity of the fabrica-
tion process to fused deposition mod-
eling (FDM) printing, the printer con-
struction could make use of many FDM
printing techniques. In fact, both the
3-D motion platform and the slicing
software originate from an FDM printer
design. However, the traditional bond-
ing approaches as demonstrated in
conventional 3-D printing technologies
is not suitable in this case. In FDM and
selective laser sintering (SLS), the bond-
ing is activated by heat. In stereolithog-
raphy, or SLA, the bonding is achieved
with ultraviolet light shining directly
onto light sensitive resin. None of these
are suitable for bonding fabric yarn. In-
stead, Hudson looked into how people
make clothes, and developed a new type
of printer head that specifically handles
yarn feeding and bonding. Material de-
livery and bonding is done with a needle
bonding mechanism—as the machine
51
feature
Figure 1. The needle felting print head.
deposits yarn along the printing path, it
uses a felting needle to pierce through
to the layer(s) below. The needle drags
down individual fibers from the yarn
below, entangling them in place.
Figure 1 shows the print head as the
heart of the felted yarn printer. At the
tip of the print head is a felting needle.
The print head drives the needle up
and down through the incoming yarn
and into the base of previously print-
ed fibers. The yarn is delivered with a
feed-lock mechanism driven by a servo
motor. A yarn guiding tube angled to
the left allows the new yarn to enter a
shared “yarn catch” section, where the
barbs on the needle can catch it on the
downward motion. In each step, the
printer moves the print head to the work
site, performs several full length felting
punches to deposit felt and bonding,
and then moves to the next spot.
Figure 2 shows the printing result: A
felt teddy bear that is soft, flexible, and
cuddly. It feels somewhat like hand-
knitted material, but in 3-D form. The
material used in this project limits
printing resolution and accuracy, but
the advantage is the felt 3-D printer of-
fers the opportunity to form 3-D arti-
facts that feel like fabric. Compared to
traditional pipelines of making soft,
fluffy toys—which require tons of know-
how to interpolate abstract 2-D shapes
and sew them into 3-D objects—felt
printing takes advantage of layer-based
3-D forming. The user can design 3-D
shapes digitally and conveniently con-
vert the design to a soft physical form.
SOFT PRINTING
WITH FABRIC SHEETS
The soft printing technique with fabric
52
yarn forms a 3-D geometry, where the
machine deposits the yarn at each of
the building step to infill the 2-D layers.
This results in a repeated punching mo-
tion at each 2-D layer, and limits the ac-
curacy and height of the model. There is
an alternative, using a 2-D fabric sheet
as the building material [2]. Different
from fabric yarn, fabric sheet has the
advantages of a pre-fabricated uniform
surface, which guarantees an even 2-D
fabric layer. With fabric sheets, the
printer does not need to deposit fabric
material additively at each of the tiny
building steps, which can potentially
achieve higher printing results.
Similar to the felt yarn 3-D printer,
a customized layered fabric printer is
developed to handle fabric sheets. The
fundamental concept is to still pile up
layers of 2-D shapes to form a 3-D shape;
but instead of building each of the 2-D
layers additively, we involve subtractive
cutting for 2-D geometry generation.
Conceptually, our fabric sheet printer is
related to the laminated object manu-
facturing (LOM) technique—we take
in sheets of material and cut out the
material to form a 2-D shape. However,
because a fabric sheet is much more
flexible than paper or metal (which are
the two materials we see in LOM print-
ing), it presents a unique set of handling
and cutting challenges. Specifically,
tool-based cutting, used in metal and
paper printers, is likely to snag and de-
form fabric layers, causing problems
for shape forming. Laser-based cutting,
though snag-free, requires layers to be
cut on a separate bed and moved to the
print in progress, potentially requiring
a complex transfer system.
Our printer design (see Figure 3) uses
laser-based cutting and solves the deliv-
Figure 2. A 3-D printed soft teddy bear: 3-D digital model rendering (left), printing
in process (middle), and the printed result (right).
ery problem using an inverted cutting
bed, which is located vertically on top
of the bonding platform where the final
result takes shape. Our design avoids
complex fabric handling schemes by
making the best use of the Z-axis mo-
tion. To start printing, the fabric is fed
into the bottom face of the inverted cut-
ting bed and held in place with a vacu-
um machine. A laser beam is generated
from a 40-watt CO2 laser tube, guided
into a print head and then pointed up
toward the inverted bed. The print head
can move freely across the X-Y platform
to cut a 2-D geometric shape out of the
felt sheet. After cutting one layer, the
print head moves out of the building
area to allow the bonding platform to
raise up along the Z-axis to catch the cut
piece. At that moment, the vacuum is
off so the 2-D shape can be transferred
naturally to the bonding platform. The
machine then lowers the bonding plat-
form. A heating disc located at the bot-
tom of the print head is then used to
bond the material. The printing proc-
ess repeats for each 2-D layer until the
3-D model is fully printed. The printed
result is a rectangular prism made up
of layers of fabric. After removing the
outside support material, the printed
object is revealed.
Figure 4 is one of the examples of a
soft, flexible Stanford bunny printed
with layers of the fabric sheet. Our
technique handles overhang shape
well (the bunny ears) and preserves the
details seen in the digital 3-D model.
SOFT PRINTING AND BEYOND
As we add fabric to the 3-D printing
material library, we can go one step
further—from exploring the use of soft
printing material to printing functional
XRDS • SPRING 2016 • VOL.22 • NO.3

Figure 3. Our layered fabric printer.
Figure 5. Beyond soft printing: A toy bear arm embedded with mechanical
structures (left); a soft starfish pendant as a touch sensor (top right); and a soft
cellphone case with an embedded coil to activate LED (bottom right).
and structural soft objects—by com-
bining other fabrication techniques
and using conductive material.
In the felted yarn printer project, sev-
eral techniques have been explored to
combine soft print patches onto hard
printed objects (PLA plastic, in our case)
using a nylon-based interface. Figure 5
is one example of a soft printed toy with
embedded components, in this case
tendon wire and a nylon mesh tube. The
toy bear’s arm is bendable and can be
activated by pulling the tendon wire.
By bridging between hard and soft ma-
terials, the combined use of both hard
and soft printing opens new opportuni-
ties to create variations of 3-D printing
results. It is not too hard to imagine
a future where a full-scale soft teddy
bear with embedded plastic eyes and
moveable mechanical parts can be fully
XRDS • SPRING 2016 • VOL.22 • NO.3
Figure 4. A 3-D printed Stanford bunny.
printed directly from a 3-D printer.
In the layered fabric printer project,
there are possibilities for embedding
electric components, such as sensors
and actuators, into soft prints. Our lay-
ered fabric printer is designed with a
second material feeder. As a result, it
supports soft printing with two differ-
ent types of fabric sheets at the same
time. By assigning one of the feeders
with a conductive fabric sheet, the print-
er produces soft circuits inside three-
dimensional geometries. Figure 5 also
showcases a starfish pendant, which
serves as a capacitive touch sensor, and
a soft cellphone case with a coil antenna
that has been printed with a conductive
fabric sheet in the middle layers of the
model. The case can harvest energy di-
rectly from the cellphone’s NFC signal
and light up an LED accordingly. These
two examples tease the potential future
of soft 3-D printing with embedded
circuits. Printed results may serve not
only visual and tactile goals, but may
also have electromechanical properties
achieved with one-off printed circuits.
CONCLUSION
By using fabric yarn and sheets, 3-D
printed results are no longer hard
and distant, but can be soft and flex-
ible. Looking forward, combining soft
prints with hard 3-D printing and em-
bedding simple electric circuits for
sensing and actuating can lead the way
to cutting-edge interactive devices. Our
approach is one important step toward
printing fully functional, but soft and
human-friendly artifacts.
References
[1] Hudson, S. E. Printing teddy bears: a technique for 3D
printing of soft interactive objects. In Proceedings
of the SIGCHI Conference on Human Factors in
Computing Systems (CHI ‘14) (Toronto, Canada, April
26-May 1). ACM, New York, 2014, 459-468.
[2] Peng, H., Mankoff, J., Hudson, S. E., and McCann J. A
layered fabric 3D printer for soft interactive objects.
In Proceedings of the SIGCHI Conference on Human
Factors in Computing Systems (CHI ‘15) (Seoul, Korea,
April 18-23). ACM, New York, 2015, 1789-1798.
Biographies
Huaishu Peng is a Ph.D. student in the Information Science
Department at Cornell University and a research associate
at Disney Research Pittsburgh. He considers himself
a maker, a tinkerer, and a designer who enjoys making
physical objects. Research-wise, he builds software
systems and machine prototypes that make the design and
fabrication of 3-D models interactive. He also looks into
new techniques that can fabricate 3-D interactive objects.
Scott Hudson is a professor in the Human-Computer
Interaction Institute within the School of Computer
Science at Carnegie Mellon University where he serves as
the founding director of the HCII PhD program. Elected
to the CHI Academy in 2006, he has published more
than 150 technical papers. He has regularly served on
program committees for the SIGCHI and UIST conferences.
He also served as a founding associate editor for ACM
Transactions on Computer Human Interaction.
Dr. Jennifer Mankoff is an associate professor in the
Human Computer Interaction Institute at Carnegie Mellon
University. She earned her B.A. at Oberlin College and
her Ph.D. in computer science at the Georgia Institute of
Technology. Her research enhances the human experience
with technology. Her goal is to combine empirical methods
with technological innovation to construct middleware
(tools and processes) that can enable the creation of
impactful applications. Most recently, her work has
focused on 3-D printing and its potential for creating
custom assistive technologies for people with disabilities.
James McCann obtained his Ph.D. in 2010 from Carnegie
Mellon University. His research hours are spent at Disney
Research Pittsburgh developing systems and interfaces
that operate in real-time and build user intuition; lately,
he has been dabbling in the creation of physical objects.
He also makes video games as TCHOW llc, including recent
releases “Rktcr” and “Rainbow.”
© 2016 Copyright held by Owner(s)/Author(s).
Publication rights licensed to ACM.
1528-4972/16/03 $15.00
53
feature

Lots of Parts,
Lots of Formats,
Lots of Headache
After three decades of digitally fabricating the world’s wildest
architecture, Zahner’s R&D team discuss trials, tribulations, and a
path to personalized production.

By James Coleman, Craig Long,

Andrew Manto, and Trygve Wastvedt
DOI:10.1145/2893501

F
actory produced architecture typically conjures images of standardization like post
WWII housing (e.g., the Lustron or Levittown housing projects) where you could
get anything you wanted just as long as it was black [1]. The history of prefabricated
architecture is diverse with varying levels of success. Each of these projects chased the
elusive trope of “more for less,” leveraging the efficiencies of systematic mass production to
provide a high level of quality for the masses. This process mainly consists of anticipating the
consumers’ needs, designing a product with a highly engineered production sequence, and
producing as many of the same products as possible.
While this is a widely utilized approach, people want options. Just as Henry Ford learned
about singular solutions for automobiles [2], this is especially true of the built environment.
The multitude of failed mass housing in design, building site, building use, plex architectural projects without an
attempts over the years is key evidence. size, local labor force, and culture. The overwhelming investment in process
Today the shape, size, and complex- delivery of architecture projects is in a engineering? To combat this prob-
ity of architectural projects can vary sense “personal production” at scale. lem, Zahner engineers found a natu-
widely. This is heightened by the grow- Neil Gershenfeld wrote, “As it turns ral partner in computer numerically
ing availability and sophistication of out, the ‘killer app’ in digital fabrica- controlled (CNC) machines. To man-
3-D modeling software used by archi- tion, as in computing, is personaliza- age the diversity of parts involved in
tects. Software that allows for bound- tion, producing products for a market complex architectural projects, digital
less geometric experimentation [3] and of one person” [4]. fabrication provided the flexibility of
helps set high expectations for the pro- The trouble with creating bespoke production that we needed.
ducers of architectural projects. projects at scale is they are cost pro-
However, unlike industries that pro- hibitive. Because economies of scale DIGITAL FABRICATION AT ZAHNER
duce large quantities of the same prod- do not exist, it is not possible to in- CNC machines became an ubiquitous
uct, architectural projects (like muse- vest equal energy into the process part of our process at Zahner in the
ums) are only produced once and often engineering of architectural projects late 1980s. Digital fabrication enabled
composed of hundreds of thousands as other industries have done. So the the linking of parametrically defined
of unique parts. These projects are problem becomes, how do we accom- computer models to our production
inherently custom due to variations modate the ever increasingly com- machines. This link allowed for the

54 XRDS • SPRING 2016 • VOL.22 • NO.3

 A view of the recently completed Peterson Automotive Museum designed by KPF Architects and fabricated by A. Zahner Co.
creation of whole families of unique
mechanical parts and assemblies. The
batch processing of dissimilar parts,
or mass customization [5], became
standard practice at Zahner.
The adoption of these technologies
made many exciting projects possible:
the Experience Music Project, the de
Young Art Museum, the Tacoma Mu-
seum of Glass, Emerson College Los
Angeles, and, most recently, The Peter-
son Automotive Museum (see Figure
1). Forms and structures never thought
possible, became achievable because
of the precision of computer control.
Since then, we have expanded to nu-
merous other kinds of computer-con-
trolled machines and robots for the
cutting and shaping of metals.
While the linking of computer mod-
els to production machines has been
foundational in our process, we have
learned many lessons over the years.
Difficulties in machine communica-
tion, programming, and hardware flex-
ibility have become increasingly bur-
Photo by A. Zahner Co.
densome, leading us to pursue ground
up digital fabrication technologies.
Although we continue to encounter
numerous problems with business as
usual digital fabrication related to ma-
XRDS • SPRING 2016 • VOL.22 • NO.3
chine integration and operation, we
believe digital fabrication can get bet-
ter from where it stands today.
The mantra that a CNC machine
cares little whether it is cutting a line,
circle, square, or spline [6] may be
true,however this does not tell the full
story. Control systems vary between ma-
chines, thus requiring specific instruc-
tions to perform a desired operation.
Common inputs for these machines
include geometry, point data, raster in-
formation, proprietary code, or G-code.
In all cases, this input eventually boils
down to motor communication signals.
While the output of all these processes is
more or less of a similar flavor, the ways
in which it is created varies widely. After
interacting with dozens of machine pro-
gramming software over the years, we
began to ask “Why all this variation, if
the end goal of each is the same?”
Take for example one of our wa-
terjet machines. It has its own CAD,
computer-aided manufacturing, and
operation software distinct from our
design engineering workflow. Data
from our master model must be trans-
lated multiple times in order to com-
municate with the waterjet. The water-
jet communication workflow diagram
(see Figure 2) describes the typical
steps necessary to fabricate a part on
the water jet from a 3-D model.
During the process, surfaces be-
come curves, curves are broken into
lines and arcs, arcs are rebuilt, quali-
ties are assigned, machining param-
eters outlined, and eventually point
data is translated into machine code.
The additional energy necessary to
integrate with these varying protocols
is arduous and does not add any value
to the project. Combined with the fact
that each machine has a unique set of
constraints, we experience a major set-
back in our process. If each operation
requires tens of minutes per part, this
adds up to several weeks of time on a
typical project. It is especially frustrat-
ing when you consider the point data
that is finally extracted already existed
in a purer form in the original model.
At Zahner we have managed this
constraint through the development
of sophisticated software automation
routines that speed up the geometry
translation and toolpathing process.
While this has, in fact, allowed us to
effectively manage, it has also proven
to be a time consuming and ineffi-
cient process. Thus outlining the key
55
feature
problems with current fabrication ma-
chines.
HUMAN VS. MACHINE
The primary offenders in our process
are proprietary software and inexten-
sible hardware, both of which make us-
ers beholden to predefined formulaic
workflows. Today, the majority of fab-
rication machines are crafted around
longevity and not flexibility, changing
little from their 1950s predecessors. A
standardized workflow in traditional
high-volume manufacturing may be
sufficient, but for non-standard pro-
ducers, like Zahner, diversity is taxed.
Production is therefore difficult for a
company with diverse output.
A primary reason behind the ex-
istence of standardized production
methods is the creation of fabrication
machines requires expertise in a wide
array of fields, such as mechanical en-
gineering, material science, automa-
tion, controls, electrical engineering,
and design computation. This makes
it is difficult for an individual or a sole
company to personalize the technol-
ogy behind digital fabrication. This
Figure 1. Diagrams of Zahner fabricated projects illustrating wide formal variations.
Figure 2. Diagram describing a typical convoluted fabrication workflow.
56
With the creation
of bespoke
fabrication methods,
Zahner is poised
to dramatically
expand what is
possible to build.
hurdle in machine design leads to
disjointed teams, where many subcon-
tractors work in isolation during prod-
uct development and produce piece-
meal machines/workflows. In fact, it
is not uncommon for machines to run
firmware that is upwards of a decade
old with antiquated user interfaces.
One must ask, “… is it really necessary
to divide the programming and ma-
chine operating functions within the
shop? Could programming, like other
tooling, be done closer to the floor or by
people on the floor?”[7].
The effects of disjointed workflows
can also be seen on the operational
side of production. Everything from
file management to equipment op-
eration varies from one machine to
the next, even if the functionality is
the same. This lack of consistency re-
quires us to have dedicated program-
mers and operators with knowledge
in each specific system. This limits
our production flexibility and propa-
gates redundancy.
PROJECT SPECIFIC MACHINES
This brings us to the question, what
would personal fabrication/produc-
tion look like at Zahner? We see it as
an agile production strategy, where
multipurpose and project specific ma-
chines are designed into our manufac-
turing process. These machines can
be there when we need them, and re-
commissioned or reconfigured when
we don’t. This would require a number
of features currently not incorporated
in digital fabrication machines. The
ability to reconfigure machine compo-
nents requires a level of modularity in
both the hardware and the software,
as well as in controls. Along with mod-
ularity, the mechanical components
must be non-specific in design to pro-
vide a range of operational metrics.
Industrial robots are a good exam-
ple of “generalist’ machines that are
multipurpose in nature. By changing
the end effector on an industrial arm,
one can quickly switch between fabri-
cation methods while maintaining the
same programming logic. While ro-
bots prove to be very useful in increas-
ing our flexibility, for many tasks they
are overly complex only gigantic robots
can achieve high stiffness values.
We plan to counter the limitations
and difficulties of industrial robots
with a complementary research tra-
jectory into reconfigurable, linear
actuators in the form of cartesian ro-
bots. Using both serial and parallel
kinematics we can quickly prototype
various motion systems by simply
bolting the actuators together in dif-
ferent orientations. This recombi-
nation of hardware is supported by
dexterity in software development
born during our many years of deploy-
XRDS • SPRING 2016 • VOL.22 • NO.3

Figure 3. Simplified workflow of a multi-purpose machine.
Figure 4. A speculative segment of the nearly endless configurations possible with
reconfigurable cartesian robots.
ing digital fabrication techniques on
diverse architectural projects. This
framework of modular hardware and
software enables the rapid prototyp-
ing of rapid prototyping machines
and project specific fabrication tech-
niques. Machines born from this
process will be a direct manifestation
of project needs without the baggage
of secondhand technologies.
Benefits of a Zahner-based system
include common programming lan-
guages, scalable production, unique
production methods, adaptable pro-
duction methods, IP development, uni-
versal operators, expanded company ex-
pertise and knowledge, and decreased
training. The ability to have operators
and programmers manage a variety of
machines increases their agency and
XRDS • SPRING 2016 • VOL.22 • NO.3
provides scheduling flexibility. This
system of agile production would break
the paradigm of dedicated operators for
a system that instead develops a univer-
sal technically advanced labor force.
Like the revolution of constraint-
based computer modeling over static
architectural drawings, multi-pur-
pose CNC machines can redefine how
architectural projects are produced.
With the creation of bespoke fabri-
cation methods, Zahner is poised
to dramatically expand what is pos-
sible to build. Project specific fabri-
cation machines will combine metal
working expertise with hardware,
software, controls, sensing, and mod-
eling innovations. We believe this
combination will lead to an ecosys-
tem of design-driven machines that
diversifies the way we make things
and circumvents the pitfalls of pre-
defined workflows.
PERSONAL FABRICATION
IN ARCHITECTURE
At Zahner, we are pursuing a fundamen-
tally different instantiation of digital
fabrication and automation. We plan to
deploy hardware, software, and a labor
production strategy that can be flexible
and interoperable to tackle a broad set
of geometries, operations, and client
requirements at volume. This flexibil-
ity will provide alternatives to one-size-
fits-all digital fabrication, allowing it to
be personalized to the unique require-
ments of architectural projects.
It is often commented by architects
and building professionals that the
construction industry is sluggish in its
adoption of new technologies. Com-
parisons are often drawn between the
building industry and the highly auto-
mated production of automobiles and
airplanes [3]. These comparisons as-
sume applicable technologies exist but
are not being incorporated. The R&D
team at Zahner see things differently.
The production of one-off architectur-
al projects requires a fundamentally
different strategy than mass produc-
tion, and thus fundamentally different
technologies. It’s time to personalize
digital fabrication, because one size
does not fit all.
References
[1] Ford, H., and Crowther, S. My Life and Work.
Doubleday, New York, 1922.
[2] Grandin, G. Fordlandia: The Rise and Fall of Henry Ford’s
Forgotten Jungle City. Metropolitan. New York, 2009.
[3] Kolarevic, B. Architecture in the Digital Age: Design
and Manufacturing. Spon Press, New York, 2003.
[4] Gershenfeld, N. A. Fab: The Coming Revolution on
Your Desktop—from personal computers to personal
fabrication. Basic Books, New York, 2005.
[5] Iwamoto, L. Digital Fabrications: Architectural and
Material Techniques. Princeton Architectural Press,
New York, 2009
[6] Pine, B. J. Mass Customization: The New Frontier in
Business Competition. Harvard Business Review
Press, Boston, 1993
[7] Noble, D. F. Forces of Production: A Social History of
Industrial Automation. Oxford University Press, New
York, 1986.
Biographies
James Coleman, Craig Long, Andrew Manto, and Trygve
Wastvedt are architects, engineers, educators, machine
designers, coders, and makers within the Research and
Development team at A. Zahner Co.
© 2016 ACM 1528-4972/16/03 $15.00
57
feature

Drowning in Triangle
Soup: The quest
for a better 3-D
printing file format
File formats for additive manufacturing are lagging behind the
capabilities of 3-D printing technology itself, and no one is doing
anything about it.

By Jesse Louis-Rosenberg
DOI: 10.1145/2893503

A
t Nervous System—a design studio that combines art, science, and technology—
we create designs that push the limits of 3-D printing hardware and software.1 By
combining generative simulations, design, and 3-D printing we create complex,
customized products (see Figure 1). Through our work developing software systems
and designs for digital fabrication, we’ve repeatedly run up against the limitations of how
geometry is defined for printing, which has caused us to start to think about and develop
volumetric alternatives.
3-D printing has experienced an ware and how we represent geometry because three points already define a
explosion in popularity in the last for 3-D printing. normal plane) and two extra miscel-
five years. Not merely in the public laneous bytes. The triangles define
eye, with the advent of home print- THE STUPID TRIANGLE LIST the surface of a three-dimensional
ers, but also in industry and scientific Chuck Hull invented stereolithog- shape, a representation referred to
research. Researchers, who had been raphy in 1986, and at the same time as a “mesh.” There is no information
toiling in obscurity for 20 years, sud- developed the first file format for 3-D to gauge which triangles are next to
denly have grants and corporate back- printing, STL, which remains the pri- which, so people commonly create
ing. However, commercially available mary format used today [1]. An STL file files that do not define a coherent
technology has advanced surprisingly is quite simple; it is a list of triangles. shape at all. Other common mesh
Image by Steve Marsel

little in the 30 years since the technol-
ogy was invented. Perhaps the area
that has developed the least is soft-
1 http://nervo.us
The triangles are defined by three
points represented by 32-bit floats.
There are a couple of idiosyncrasies:
Each triangle also has a normal di-
rection (which is almost never used
formats define shapes with vertex in-
dices, supplying topological connec-
tivity information and reducing file
size. Triangles can be inside out, form
non-manifold surfaces, or simply not

58 XRDS • SPRING 2016 • VOL.22 • NO.3

Photo Credit TK

XRDS • SPRING 2016 • VOL.22 • NO.3 59

feature
Figure 1. The Kinematics Dress by Nervous System is composed of thousands of
unique interlocking components. Each dress is 3-D printed as a single folded
piece and requires no assembly.
(a) (b)
Figure 2. 2-D equivalents of different representation methods for volumes:
(a) boundary surface, (b) voxels, and (c) functional representation.
be closed at all. The format is inexten-
sible, providing no way to include ad-
ditional semantic information.
The main problem is the core idea
of a boundary surface representa-
tion, which defines a volume by its
surrounding surface. This type of
representation causes all sorts of
headaches: the aforementioned open
surfaces, overlapping shapes, gnarly
edge cases, etc. There is a more criti-
cal concern, 3-D printers create vol-
umes but our file format only defines
surfaces. What happens inside the
surface is just as important as the ex-
terior skin. An emerging area of 3-D
printing is multi-material printing.
By specifying materials with differ-
ent properties in complex configura-
60
f(v) = |v-p| - r
r software developers and 3-D printer
p Voxels are the extension of pixels to
tions, we can create meta-materials
with properties that could not exist
before. Let us take a relatively simple
example from nature; a squid’s mouth
has a gradation of stiffness that al-
lows its hard and sharp beak to attach
to its gelatinous body without tearing
it apart [2]. Currently, printing gradi-
ents of material requires either labo-
rious workarounds or direct access to
the firmware or special APIs. To ad-
vance 3-D printing technology it will
be necessary to have a universal for-
mat for specifying material properties
that can change throughout a volume.
ALTERNATIVES: FUNCTIONAL
REPRESENTATION AND VOXELS
There are two primary candidates for
volumetric representations that could
replace meshes: functional represen-
tation (f-reps) and voxels (see Figure
2). F-reps are essentially implicit sur-
face equations: Define a function that
takes a point and returns whether or
not it is inside a volume. In some ways,
this is the most basic and fundamen-
tal way to define a volume mathemati-
cally. Extending this idea slightly, we
can have a function that returns color,
density, Young’s modulus, or other
material properties. F-reps are incred-
ibly expressive and can create com-
plex shapes with more compression
than any other method. However, that
expressivity is also their greatest flaw.
Because you can describe anything
that you can define mathematically, it
is difficult to create a cross-platform
standard. How would existing CAD
packages, which use boundary sur-
face representations, read or write
such a format? You can create a f-
rep of a boundary surface, but that
would essentially be a wrapper for the
boundary representation itself and no
better. Truly utilizing an f-rep format
requires an exclusively f-rep workflow,
which creates a barrier to getting CAD
manufacturers on board.
3-D. We unitize space to a certain res-
olution and then create a rectilinear
grid of values that contain material
Working with
voxels gives us
the potential
to take advantage
of new 3-D printing
technology
and frees us
Figure 1 Photos by Steve Marsel
from many
of the headaches
of traditional
boundary
representations.
XRDS • SPRING 2016 • VOL.22 • NO.3
 properties. The primary advantage of have recently formed to bring togeth- zip file with a list of PNG images for
this method is its simplicity. It is easy er partners from different parts of the each layer and a JSON file with print
to understand, read, and write and industry to draft new standards [3]. settings. Shapeways, a 3-D printing
there is no ambiguity as to what mate- However, both of these have resulted service bureau, has begun accepting
rial occupies a position in space. in what is basically an XML wrapper similar files—zipped PNG stacks—
The problem with voxels is size. around an STL with some extra fea- which still have to be converted to
Let’s assume you want to represent tures for specifying materials. My STLs to interface with their printers.
your voxel space at the resolution of personal loathing of XML-based for- While it is encouraging to see compa-
your printer, around 50 microns is mats aside, this does little to advance nies adopt voxel formats, it is doubt-
reasonably fine, and each voxel takes the technology. In theory, it is exten- ful a solution like a zip file of images
up 1 byte, which is probably an un- sible and can eventually incorporate could become standard. First, it is
derestimate for a complex material. other data types, like voxels, but this not extensible. What happens when
That would mean even a relatively shifts the same issues from file for- there is a five-channel printer, like
small print area of a 128mm cube mats now to standards compliance in RGBA plus hardness? Second, it does
would take up 15.625GB. Even if you the future. not encourage adoption in software
compress the file on disk, it quickly There are companies incorpo- development. Reading and writing
becomes a problem to work within rating voxel formats currently. Au- such a format feels like a strange
the memory of most desktop comput- todesk’s Ember printer is the first workaround that most CAD software
ers, and there are many printers with commercially available printer that is unlikely to implement. As stated
higher resolutions and larger build can directly print voxel data. The for- previously, working with the raw, un-
volumes than this example. For this mat is a bit of a hack though. It is a compressed voxels becomes imprac-
reason, people have created hierarchi-
cal voxel formats; the most common Figure 3. (a) A diamond engagement ring 3-D printed in wax and cast
being an oct-tree, a cascade of smaller in white gold. (b). A 3-D lamp with an LED fixture.
and smaller cubes where each level
gets divided into eight equally-sized
smaller cubes. In practice, you do not
need to specify the material at the
printer’s resolution everywhere. Most
of the space will either be empty or
uniform with changes only occurring
in a fraction of the volume. A hierar-
chical voxel format only uses the fin-
est resolution where necessary. This
compression comes at the expense of
some of the simplicity that we value
in voxels, hindering the creation of
a standard that is easy to implement
and performant enough for diverse
applications. Additionally, some 3-D
printer manufacturers are concerned (a) (b)
with the possibility of someone want-
ing to specify every single point of a
printer’s volume with a different ma- Figure 4. A 3-D printed sculpture defined Figure 5. New Balance’s 3D-printed
terial. With a hierarchical format, by 1.67M lines of varying thickness. midsoles with customized cushioning.
that would be even larger than simple
vanilla voxels.

WHAT IS BEING DONE

People in the industry have known for
Figure 5 Photo Courtesy of New Balance

a long time that STL is an insufficient

file format. However, no one has tak-
en the lead in developing new stan-
dards. 3-D printer manufacturers do
not want to make their machines run
on files that no CAD software can cre-
ate, and software developers do not
want to produce files that no one can
use. Two consortiums, AMF and 3MF,

XRDS • SPRING 2016 • VOL.22 • NO.3 61

feature
tical without significant advances in
system memory.
LOOKING OUTWARD
FOR INSPIRATION
New ideas in volumetric representa-
tions are emerging out of other in-
dustries like game development and
visual effects. The sandbox game
“Dreams” by Media Molecule allows
users to create their own 3-D environ-
ments and characters based on volu-
metric tools [4]. Their system stores
models as a list of constructive solid
geometry (CSG) operations assem-
bling primitive shapes with some
f-rep-like effects. It then turns those
operations into a signed distance
function in real time.
OpenVDB is a library for voxel data
developed by Dreamworks for the
film industry [5]. VDB is a hierarchi-
cal voxel format based on B+ trees,
which are often used in databases. It
overcomes many of the shortcomings
of traditional hierarchical voxels for-
mats, allowing for arbitrary bounds
and fast random and sequential ac-
cess. Though the format is complex,
the OpenVDB library provides a cross-
platform, open-source implementa-
tion with functionality that makes it
easy to incorporate into existing work-
flows, including conversion to and
from meshes. It is extensible and al-
lows for embedding metadata. While
it is missing some functionality that
might be desired for 3-D printing, it
provides a solid foundation to develop
from, or at least a strong example of,
what other industries are doing.
UTILITY OF VOXELS
At Nervous System, we have been de-
veloping in-house tools to work with
voxel data based on the OpenVDB li-
brary. Voxels not only provide a truly
volumetric representation for 3-D
printing, but afford many other ad-
vantages compared to boundary rep-
resentations. This tool has quickly be-
come integral to the projects we work
on. Voxels allow for simple and fast
Boolean operations to combine mul-
tiple shapes. These operations are of-
ten the Achilles’ heel of complex mesh
shapes, and almost no CAD software
can do them robustly or quickly. In
voxel space, a union is simply an OR
62
By specifying
materials with
different properties
in complex
configurations,
we can create
meta-materials
with properties
that could not
exist before.
operation on each element of an array.
We use this to add hardware attach-
ments to the organic shapes we de-
sign, like inserting gemstone settings
into an engagement ring or fixtures
into a lamp (see Figure 3).
Similarly, voxels provide an ef-
ficient format for performing CSG
operations. This is a way of defining
complex shapes by combining many
simple, easily described shapes. Often
times, a structure we design will be
described as a dense network of lines
with varying thicknesses. Turning
this into a 3-D shape involves combin-
ing hundreds of thousands of cylin-
ders. This process used to take us sev-
eral minutes, but now using VDB only
requires a few seconds, allowing for
rapid design iteration (see Figure 4).
The OpenVDB library also provides
functionality for level-set operations.
Level sets are a generalization of vox-
els where each data point contains
the signed distance to the surface, so
negative values are inside and positive
are outside. This allows for defining
a shape with much higher accuracy
than the resolution of the voxel space.
We can use linear interpolation to
find the zero crossings of this field
to define our surface. It also makes
doing surface offsets easy, another
operation that is impractical for com-
plex shapes using boundary represen-
tations. One simply adds the offset
distance to each value of the level set.
We have been using this to create thin
molds of some of our designs.
Hierarchical voxel spaces can pro-
vide utilities outside of direct geometric
modeling as well. They form the basis
of adaptive numerical simulations for
fluids, crystal growth, finite elements,
etc. It also serves as a structure for per-
forming intersection tests, finding if
any point in space is inside our surface.
With a boundary surface, we have to
make a separate data structure to accel-
erate these queries—like an AABB tree,
which is slow and error prone. With
VDB, the data itself is the acceleration
structure. All of these techniques have
helped us develop 3-D printed midsoles
for New Balance sneakers, which have
a cellular structure adapted to a user’s
running data (see Figure 5).
Working with voxels gives us the
potential to take advantage of new
3-D printing technology and frees us
from many of the headaches of tradi-
tional boundary representations. We
use it to perform demanding geomet-
ric operations that would be impos-
sible with existing CAD software. I do
not know if OpenVDB is the correct
solution for a 3-D printing file format
moving forward, but for me it is shows
a format should not just solve techni-
cal issues or appease many parties. It
should be a tool that inspires people
to use 3-D printing to its full potential.
References
[1] Hull, C.W. Apparatus for Production of Three-
Dimensional Objects by Stereolithography. Patent
US4575330. 11 Mar. 1986.
[2] Miserez, A., Schneberk, T., Sun, C., Zok, F. W. and
Waite, J. H. The transition from stiff to compliant
materials in squid beaks. Science 319, 5871 (2008),
1816-19.
[3] ASTM ISO / ASTM52915-13, Standard Specification
for Additive Manufacturing File Format
(AMF) Version 1.1, ASTM International, West
Conshohocken, PA, 2013. http://www.astm.org/
[4] Evans, A. Learning from failure: A survey of promising,
unconventional and mostly abandoned renderers
for ‘Dreams PS4’, a geometrically dense, painterly
UGC game. SIGGRAPH 2015. Advances in Real-Time
Rendering, Part II. 2015.
[5] Museth, K. VDB: High-resolution sparse volumes with
dynamic topology. ACM Transactions of Graphics
(TOG) 32, 3 (2013), 27.
Biography
Jesse Louis-Rosenberg is the co-founder and chief
science officer at Nervous System. He studied math at
MIT and worked at Gehry Technologies in building design
automation. Founded in 2007, Nervous System has
pioneered the application of new technologies in design.
Their work has been featured in numerous publications
including WIRED and the New York Times, and is part of the
permanent collection of Museum of Modern Art and the
Cooper-Hewitt Smithsonian Design Museum.
© 2016 Copyright held by Owner(s)/Author(s).
Publication rights licensed to ACM.
1528-4972/16/03 $15.00
XRDS • SPRING 2016 • VOL.22 • NO.3
ACM Books.
In-depth.
Innovative.
Insightful.
The VR Book: Human-Centered
Design for Virtual Reality
By Jason Jerald, PhD
The VR Book focuses on human-centered design of virtual
reality (VR) experiences. Creating compelling VR experi-
ences is an incredibly complex challenge. When VR is done
well, these experiences can be brilliant and pleasurable, but
when done badly, they can result in frustration and
sickness. While there are many causes of bad VR such as the
limitations of technology, much is centered on a lack of
understanding of human perception, interaction, design
principles, and real users.
This book focuses on human elements of VR, such as how
users perceive and intuitively interact with various forms of
reality, causes of VR sickness, creating useful and pleasing
For more info please visit
content, and how to design and iterate upon effective VR
applications. It is not just for VR designers; it is for the entire http://books.acm.org
team, as they all should understand the basics of perception, or contact ACM at
VR sickness, interaction, content creation, and iterative design. ACMbooks-Info@acm.org
Good VR design requires strong communication between
human and machine, indicating what interactions are
possible, what is currently occurring, and what is about to M
occur. A human-centered design principle, like lean
methods, is to avoid completely defining the problem at
&C
Association for Computing Machinery
the start and to iterate upon repeated approximations and 2 Penn Plaza, Suite 701
modifications through rapid tests of ideas with real users. New York, NY 10121-0701, USA
Phone: +1-212-626-0658
Thus, The VR Book is intended as a foundation for anyone Email: acmbooks-info@acm.org
and everyone involved in creating VR experiences
including: designers, managers, programmers, artists,
M
psychologists, engineers, students, educators, and user
experience professionals. &C
Morgan & Claypool Publishers
Paperback ISBN: 9781970001129  Hardcover ISBN: 9781970001150 1210 Fifth Avenue, Suite 250
Ebook ISBN: 9781970001136  DOI: 10.1145/2792790 San Rafael, CA 94901, USA
Phone: +1-415-462-0004
Email: info@morganclaypool.com
feature

Island Three Revisited:

O’Neill cylinders and
digital materials
Huge, habitable structures in space are a staple of science fiction, but
digital materials could make them a reality.

By Daniel Cellucci and Kenneth C. Cheung

DOI:10.1145/2893505

I
n the early 1970s, a Princeton physics professor by the name of Gerard K. O’Neill started
working on what he initially thought of as a joke: A set of back-of-the-envelope calculations
regarding the design and construction of orbital habitats. These habitats used no new
sources of propulsion and no magic materials; everything was limited to the scientific
and technical understanding of his time. They were to be composed of two counter-rotating
cylinders, such that those standing on the interior surface would feel Earth’s gravity and sea-
level atmospheric pressure. The maximum size of these structures turned out to be surprising;
using steel cables like those employed in large suspension bridges, O’Neill proposed a habitat
called “Island Three,” which would measure 6.5 kilometers in diameter and 32 kilometers
in length and provide 323 square kilometers of livable space. Structures of this type would
become known as “O’Neill cylinders.”
His vision for these cylinders was posed to be, a nature preserve, a mu- The plan proposed by this group of
not small. When we talk about sus- seum to our humble past. researchers described a series of pro-
tainability, we typically refer to the The O’Neill cylinder culminated in gressively larger structures orbiting a
behaviors and technologies that al- a conference in 1975 centered around stable island in the Earth-Moon gravi-
low us to use Earth’s resources more the development of a space manufac- tational system, the L4/L5 Lagrangian
Painting by Rick Guidice courtesy of NASA.

intelligently—recycling, for instance,
instead of landfilling, or solar arrays
and wind farms instead of coal plants.
These form a worthwhile and integral
part of our future, but in his habitats
O’Neill saw a different kind of sustain-
ability: A civilization that no longer re-
lied on Earth and its fragile biosphere
for everything it needs. Where Earth
could become what it always was sup-
turing facility [1]. The backgrounds
of the presenters at this conference
reflected the diverse array of chal-
lenges in building a habitable struc-
ture of this size: geologists assessing
the various mineral sources for the
cylinders, social scientists discussing
governance in an orbital habitat, and
agronomists introducing methods for
producing food for the inhabitants.
points. Within 50 years, they predict-
ed, most of humanity would be living
in these colonies, enjoying (among
other things) beautiful views of space,
idyllic natural environments, and ze-
ro-gravity sports.
So, what happened? Well, in the
Figure 1: An illustration of Island Three.

64 XRDS • SPRING 2016 • VOL.22 • NO.3

Photo Credit TK

XRDS • SPRING 2016 • VOL.22 • NO.3 65

feature
Figure 2: Then and “Now.” The left image shows workers constructing Britain’s R101 airship. The right image is a painting that
envisions the construction of an orbital habitat. Humans were going to assemble the pieces of the structure by hand, much like
the dirigibles of the early 20th century.
40 years since its inception, we have
learned a few things that have made
it clear that this endeavor is far more
complex than we initially thought. In
the case of Biosphere 2, for instance,
we realized how difficult it is to make
stable, isolated simulacra of Earth’s
ecosystems. Unforeseen imbalances
with the inputs into the complex, cy-
bernetic life support system result in
runaway conditions and loss of the sys-
tem [2]. In addition, while the Interna-
tional Space Station is a fine example
of small teams of people working to-
gether in an isolated environment, we
have a long way to go before hundreds
of individuals can form societies that
allow them to thrive in these remote
environments. These problems are not
insurmountable, but they are wicked,
and continue to demand (and have re-
ceived) study from specialists in a wide
range of fields.
Until recently, however, we didn’t
even have a tractable idea of how to
build these structures. O’Neill’s origi-
nal calculations used the theoretical
tensile limit of steel to estimate the
maximum size of the cylinders, but
his suggestion to use multiple wound
66
cables like those in suspension bridges
focused more on finding and process-
ing raw materials for the structure
than assembling it. Nowhere, for in-
stance, in his seminal article or in the
conference that followed is the term
“robotics” mentioned. Instead, he pro-
posed 2,000 people would go into orbit,
laboriously constructing the first habi-
tats by hand, much like the dirigibles
of the early 20th century.
What we’ve learned in the interven-
ing 40 years, with experiments per-
formed both in orbit and on Earth, is
humans are not particularly efficient
builders in space. Studies in simulated
microgravity environments show them
assembling at a less than optimal 100
seconds per strut per person [3], the lack
of a consistent gravitational force ap-
pears to cause a whole host of medical
problems, and the constant bombard-
ment by both solar and cosmic radia-
tion increases the risk of cancer. Once
a sufficiently large structure is created,
however, these problems are greatly re-
duced. They won’t need bulky suits in a
pressurized environment, centrifugal
gravity on the inside of these cylinders
can approximate Earth’s 1-G, and the
structure itself can provide sufficient
radiation shielding. Using humans to
construct an environment with Earth-
like gravity and atmospheric pressure
therefore becomes a chicken and egg
problem. In order to build a human-
habitable space, we would need anoth-
er human-habitable space to house the
humans during the months they spend
building it.
Other options, such as planetary
Photo courtesy of the National Archives UK. Painting by Don Davis courtesy of NASA.
colonization, aren’t much better. The
current favorite destination, Mars,
doesn’t provide much in the way of at-
mosphere, gravity, or radiation protec-
tion. And the nine-month journey will
require much of the same infrastruc-
ture in habitat design in order to safely
transport human astronauts. More-
over, once they reach their planetary
destination, colonists will then find
themselves at the bottom of another
gravitational well similar in magni-
tude to the one they just spent tera-
joules of energy escaping on Earth,
making it that much harder for them
to return in the event of a catastrophe.
So if we decide to colonize planets,
how do we construct human habitats in
space without subjecting the first wave
XRDS • SPRING 2016 • VOL.22 • NO.3
of humans who live there to danger-
ous living conditions? Two recent ad-
vances offer some indication. The first
has already caused a recent explosion
in researchers exploring the material
properties frontier, in particular with a
competition to produce the “strongest,
lightest material.” They’ve been able to
do this by applying a set of design prin-
ciples for a class of materials called cel-
lular solids [4]. These principles show
if one starts with a solid block of mate-
rial and removes mass in a specific way,
the resulting structure behaves like the
bulk material with characteristics such
as high stiffness, strength, and fracture
toughness that are directly proportional
to the relative density of the structure to
the original solid. Researchers have used
these principles, as well as advanced ad-
ditive manufacturing techniques, to
produce small, light lattices—essen-
tially, mechanical crystals—that behave
with surprising properties. But the tech-
niques they use to make these lattices
limits their scale. Additive manufactur-
ing is energy intensive, time consuming,
and the maximum size of the objects it
can create is limited.
One feature of cellular solids theory
is many desirable properties are scale-
invariant. The only requirement is for
the “cells” that compose a solid be sized
on the order of 10 to fit in the smallest
dimension of the overall structure. The
dimensions of these cells can be 10 mi-
crons or 10 meters, as long as there are
a minimum of 10 of them in any direc-
tion, the resulting solid will behave like
a bulk material with properties that
depend on the relative density. It’s with
this insight that the second advance—
digital materials—becomes important
for space construction.
The idea of digital materials is this.
Start with a part—a simple piece of car-
bon fiber, aluminum, or even cast basalt
from melted lunar regolith—that has
been made with a machine whose only
job is to fabricate this one object. This
part should be designed so it can inter-
lock with a few others to form a three-di-
mensional cell—a single piece of a crys-
tal. With a mechanism that can perform
this crosslinking reliably, and a robot
that can climb on this crystal, we have
the means to assemble enough parts to
reach kilometer-scale construction in
space with minimal human assistance.
XRDS • SPRING 2016 • VOL.22 • NO.3
In addition, the size of the robot and the
machine that makes the parts is deter-
mined by the part size, not the size of
the structure, so we can pick a part size
that is appropriate for our manufactur-
ing process. We can build structures
orders of magnitude larger than our fac-
tory, simply because the partially com-
plete structure becomes the platform
on which the rest of it is built.
The mechanical connection also
adds a few interesting capabilities.
What if a few hundred of the parts in
our growing crystalline home are later
crushed by a colliding asteroid? The
same robots can simply undo the mech-
anisms for the parts surrounding the
failure, and replace the broken parts
with new ones. Or maybe a structure
built from digital materials is no longer
necessary—it was once a tank for hold-
ing fuel in a launch vehicle, but has en-
tered into orbit and is now empty mass.
Instead of jettisoning the mass that was
so expensive to launch, it can be broken
down and the component parts can be
reused for something else.
With these two insights—digital
materials and cellular solids—we can
see what space colonization might look
like on the road to Island Three. We
wouldn’t have to start large. Just 40 me-
ters in diameter is sufficient for humans
to comfortably function in a centrifu-
gal, Earth-like gravity environment and
not be excessively disoriented by the
Coriolis force [5]. If such a habitat were
100 meters long, it would still have less
than half the mass of the International
Space Station and would provide nearly
175 times the pressurized volume.
Even at this relatively small scale,
digital materials can play a vital role.
Of the mass sent into orbit by every
Delta IV rocket in the form of the sec-
ond stage vehicle and the fairing [6],
about four metric tons is wasted. If
this precious mass, which would have
been discarded as soon as the main
payload entered its target orbit, were
instead kept and recycled into build-
ing a larger structure, we could reduce
the numbers of launches and make the
effective cost that much lower. Once
the base habitat is built, it can be con-
stantly augmented and maintained
with material sources that don’t have
to come from Earth. The simplifica-
tion of the structure to a few part types
means we can optimize automated
manufacturing processes to create
components from the resources avail-
able on the Moon or elsewhere.
This toehold in space would pro-
vide a laboratory where we could safe-
ly work on the problems of colonizing
the solar system. Here, we can figure
out how to grow gardens in space,
how to maintain ecosystems far from
Earth, and how to make environments
that can support humans. Digital ma-
terials give us the blueprint for the
platform—the means to design and
build these large structures. But, as
O’Neill noted 40 years ago, “This work
should be discussed and debated as
widely as possible, by people with a
range of technical and artistic talents,
and by people who claim no special
talent beyond the ability to work hard
for a worthwhile goal.”
References
[1] O’Neill, G. K. The colonization of space. Physics
Today. 27, 9 (1974), 32-40.
[2] Severinghaus, Jeffrey P., et al. Oxygen loss
in Biosphere 2. EOS, Transactions American
Geophysical Union 75, 3 (1994), 33-37.
[3] Lake, M. S. et al. Evaluation of hardware and procedures
for astronaut assembly and tepair of large precision
Reflectors. NASA/TP-2000-210317. Scientific and
Technical Information Office NASA. 2000.
[4] Fleck, N. A., Deshpande, V. S., and Ashby, M. F.
Micro-architectured materials: Past, present and
future. Proceedings of the Royal Society of London
A: Mathematical, Physical and Engineering Sciences.
466, 2121. The Royal Society, 2010.
[5] Stone Jr., R. W. An overview of artificial gravity. In
Fifth Symposium on the Role of the Vestibular Organs
in Space Exploration. NASA-SP-314. Scientific and
Technical Information Office. NASA, Washington D.C.,
1973, 23–33.
[6] United Launch Alliance. DELTA IV Payload Planners
Guide. Sept. 2007.
Biographies
Daniel Cellucci is a Ph. D. student studying mechanical
engineering at Cornell University, and a researcher with
the Coded Structures Lab at NASA Ames Research Center.
His work explores the robotic assembly of large-scale
space structures, as well as the impact of digital materials
on the future of space exploration. He is the recipient of
the NASA Space Technology Research Fellowship.
Kenneth C. Cheung is a research scientist at the National
Aeronautics and Space Administration (NASA) Ames
Research Center (ARC), conducting research at the
intersection of design, mechanical engineering, and
material science. His current laboratory work focuses
on applying building-block based (digital) materials
and algorithms to aeronautical and space applications.
He also serves as an advisor to the NASA ARC Chief
Technologist (CCT) on matters concerning materials
science and engineering, rapid prototyping, and advanced
manufacturing. The NASA ARC CCT office helps to
identify, define, develop, and integrate new and emerging
technologies for application to NASA and national goals
through the NASA Space Technology Mission Directorate.
© 2016 Copyright held by Owner(s)/Author(s).
Publication rights licensed to ACM.
1528-4972/16/03 $15.00
67
feature

3-D Printing:
Green or not?
3-D printing could herald new advances in sustainable production,
that is, so long as it does not become a sustainability hazard itself.

By David Rejeski
DOI: 10.1145/2893507

O
ver the next 15 years, an estimated three billion new consumers from developing
countries will enter the middles class, a trend likely to increase the consumption
of both materials and energy with a concomitant growth in waste production.
Though not a panacea, 3-D printing could become part of a larger strategy to close
material loops, extend product life, increase energy efficiency and value extraction, and
reduce supply and production footprints [1]. The increasing growth of 3-D printing in the
marketplace provides a rare opportu- environmentally benign, and energy tion of 3-D printers, including material
nity to minimize energy, health, and efficient; research that is necessary to (polymers, metals, ceramics), feedstock
environmental impacts as the technol- create a better model for sustainable form (powder, filament, liquid, sheet),
ogy is developed and commercialized production. As David Dornfeld of the process (from extrusion to bed fusion
rather than after the fact. As a poten- University of California-Berkeley has and directed energy deposition), use lo-
tially transformational technology, 3-D noted, “3-D printing is an innovation cation (industrial facility, maker space,
printing could be used in the service of that will change the way manufactur- or home), and post-processing options.
sustainability. But as we have seen with ing is practiced in a wide range of appli- In addition, new technologies are com-
other disruptive technologies—from cations. The challenge will be to insure ing online, such as continuous liquid
nanotechnology to synthetic biology— that the environmental impacts of this interface production (CLIP), along with
there is no guarantee that this will hap- new technology are fairly evaluated as novel application areas such as bio-
pen. Pessimistic observers have noted we rush toward this ‘next big thing’” [2]. printing [5]. A fuller understanding of
3-D printing could instead, “herald the Much of the scientific expertise the impacts will require more research
apotheosis of consumerism, instant needed to do research on sustainable on lifecycle assessment, especially the
gratification, the throwaway society” [2]. 3-D printing already exists in estab- supply-chain footprint; occupational
3-D printer shipments are forecast lished fields. There is also interest from health issues, including exposure as-
to more than double every year from researchers in areas such as industrial sessment and control; and energy use,
2016 to 2019, by which time worldwide ecology and environmental manage- especially studies that compare addi-
shipments are expected to reach more ment, but funding from U.S. agencies, tive manufacturing to conventional
than 5.6 million [3]. The value of the 3-D like the National Science Foundation processes and explore “embedded” en-
printing market is projected to grow to and the Department of Energy, has been ergy demands behind different mate-
$16.2 billion in 2018 (from $288 million minimal to date. Assessing the impacts rial choices. Finally, more research is
in 2012) [4]. Already, the use of 3-D print- of 3-D printing will be neither simple needed on end-of-life waste and its im-
ers has far outstripped the research nor inexpensive due to the increasing pacts [6]. A recent study that examined
needed to ensure the technology is safe, variety of options affecting produc- the toxicity of 3-D printed parts called

68 XRDS • SPRING 2016 • VOL.22 • NO.3

 attention to the “need for strategies
for the safe disposal of both parts and
printer waste materials” [7].
As we wait for such research to ma-
terialize, more 3-D printers are being
poured into the marketplace. However,
there are opportunities for users and
producers of 3-D printers to apply exist-
ing research combined with best prac-
tices from occupational health and safe-
ty and environmental management.
Though it has yet to engage with 3-D
printing, the United States Environmen-
tal Protection Agency (EPA) is poised
to address the issue of sustainability
in 3-D printing. The EPA has expertise
developed through its Design for Envi-
ronment program and years of work in
“green chemistry.”
A few years ago, Robert Olson, a se-
nior fellow at the Institute for Alterna-
tive Futures, proposed the following set
of provisional principles for 3-D print-
ing based on green design principles.
It should be noted the principles (repro-
duced here in their entirety) have yet to
be tested and adapted:
˲˲ First, consider the sustainability of
IImage by Liesbeth Mobach
3-D printer production. Develop manu-
facturing processes that use energy and
materials efficiently, use renewable re-
sources where possible, and treat waste
as a design failure.
˲˲ Second, design 3-D printers for en-
XRDS • SPRING 2016 • VOL.22 • NO.3
ergy and resource efficiency. Strive for
more energy efficient operations, pre-
vent waste, improve durability, and al-
low easy repair and upgrading.
˲˲ Third, use feedstocks that are re-
newable and bio-degradable whenever
technically and economically practica-
ble. When that is not possible, use eas-
ily recyclable feedstocks and insure they
are properly recycled.
˲˲ Fourth, design both 3-D printers
and feedstocks for safe and healthy op-
eration. Minimize dangers of exposure
to toxic materials of any kind.
˲˲ Fifth, provide product take-back
and recycling of 3-D printers, feedstock
cartridges, and other components and
recycling of discarded printer products.
˲˲ Sixth, provide easy-to-understand
information on safe operation, mini-
mizing particle exposure, using fea-
tures that improve energy efficiency and
minimize waste, and doing cleaning
and other maintenance needed to mini-
mize environmental impacts. Promote
sustainable consumption over frivolous
use. [2]
The existing state of affairs provides
an opportunity for students, early ca-
reer investigators, and engaged “mak-
ers” to create a community of practice
dedicated to the development, testing,
and adaption of “green” principles
for 3-D printing. The creation of such
a community could help to identify
knowledge gaps, research needs, and
areas of consensus that could form the
basis for more formal codes of practice.
Green is not the default value of new
technologies. We need to make a con-
scious effort to make environmental
stewardship happen.
References
[1] Nguyen, H., Stuchtey, M., and Zils, M. Remaking the
industrial economy. McKinsey Quarterly. February 2014.
[2] Olson, R. 3-D printing: A boon or a bane? The
Environmental Forum 30, 6 (2013) 34-38.
[3] Gartner. Gartner says worldwide shipments of 3D
printers to reach more than 490,000 in 2016. Press
release. September 29, 2015. http://www.gartner.
com/newsroom/id/3139118
[4] Canalys Inc. 3D printing market to grow to 16.2
billion USD in 2018. March 31, 2014. Press release.
http://www.canalys.com/newsroom/3d-printing-
market-grow-us162-billion-2018
[5] Tumbleston, J., et al. Continuous liquid interface
production of 3D objects. Science 347, 6228 (2015),
1349-1352.
[6] Rejeski, D. and Huang, Y. Environmental and
Health Impacts of Additive Manufacturing: An NSF
Workshop Report. Woodrow Wilson Center. 2015.
[7] Oskui, S.M., et al. Assessing and Reducing the
Toxicity of 3-D Printed Parts. Environmental Science
and Technology Letters 3, 1 (2016) 1-6.
Biography
David Rejeski directs the Science and Technology Program
(STIP) at the Wilson Center, a non-partisan think tank in
Washington, DC. STIP focuses on emerging technologies
and the challenges and opportunities they present for
public policy.
© 2016 Copyright held by Owner(s)/Author(s).
Publication rights licensed to ACM.
1528-4972/16/03 $15.00
69
feature

Fabrication Lends
a Hand: Creating
custom assistive
technology
Lost your hand in a lightsaber fight? No problem, we can fix that.
Rapid and consumer-grade fabrication tools could revolutionize
the way we design and deliver assistive technologies.

By Erin Buehler
DOI: 10.1145/2893497

W
e’re still a few years out from an off-the-shelf, brain-computer-interfaced,
fully articulated prosthetic hand a la Luke Skywalker at the end of “The Empire
Strikes Back.” However, we have cracked the problem of highly customized
and inexpensive prosthetics. A 3-D printed prosthetic hand is one example
of the many types of assistive devices that can be created with rapid fabrication
technologies. With consumer-grade tools, we can make bespoke assistive devices that are
robust and affordable. If we work to make fabrication and design tools more accessible,
we can empower people with disabilities to design assistive devices for themselves.
For the last few years, I have been (DIY) practices to create better assis- to complete a task that they would oth-
doing research on both the applica- tive technology solutions. Assistive erwise be unable to do. Try to think
tions of fabrication tools and their ac- technology demands a lot of flexibil- of an assistive device—go ahead, I’ll
cessibility. I’ve spent time in the field ity and customization, both of which wait. If you’re interested in web or in-
working with special education teach- have been fulfilled over the years with terface design, you might be thinking
Image by Belekekin / Shutterstock.com

ers, therapists, and students with dis-
abilities trying to find out how we can
maximize the use of these tools and
move beyond 3D-printed key chains
and laser-cut, finger-jointed boxes.
The accessibility community is a
prime space for the contributions of
fabrication, as they have always been
engaged in making and do-it-yourself
everything from floral tape to thermo-
formed plastics. It’s only logical that
tools such as 3-D printers, laser cut-
ters, and CNC mills find a home here.
UNDERSTANDING
ASSISTIVE TECHNOLOGY
Assistive technology is any technology
that enables a person with a disability
of screen readers that use embedded
markers within text to navigate infor-
mation and provide audio output. Or
maybe you’re a hardware person, so
onboard sensors for obstacle detec-
tion and avoidance systems as an op-
tion for power wheelchair users come
to mind. But if you stretch your inter-
nal definition just a bit, you’ll realize

70 XRDS • SPRING 2016 • VOL.22 • NO.3

Photo Credit TK

XRDS • SPRING 2016 • VOL.22 • NO.3 71

feature

even some of the simplest objects are

examples of assistive technology. The
Design tools for U.S. With a 3-D printer, you can fabri-
cate a mobile device mount and fit it
widely recognizable white cane is an rapid fabrication to any wheelchair for about $5. This
assistive technology that provides tac-
tile and audio feedback for navigation
present challenges of means you can create a simple piece
of assistive technology for 5 percent
to a user with a visual impairment. accessibility relevant of the cost of an off-the-shelf product.
Perhaps you’re asking yourself,
“Well, if that’s the case, then couldn’t
to novice designers Customization. An assistive de-
vice can be a great thing—it can em-
almost anything be an assistive tech- and persons with power a person to be independent, but
nology?” The answer is: Yeah, prob-
ably. (We’ll talk more about this later.)
disabilities. this only works if the user adopts the
technology. Assistive tech has a long
“So what,” you say and sip your latte history of user abandonment [2]. The
of indifference, “Why do I care about device has to meet certain criteria for
this? I just like fabrication tools.” the end user to stick with it. It needs to
Here’s what: Fabrication tools lend accomplish the task it’s designed for,
themselves really well to two big bar- into (e.g., a developmental disability be relatively easy to use or master, and
riers to the success of assistive tech- like Down Syndrome), you can age into it needs to be a reasonable price. But
nologies—cost and customization. (e.g., cognitive impairments related to perhaps most importantly, an assis-
$$$ a.k.a Cost. Assistive technol- dementia or Alzheimer’s), or you can tive technology device must appeal to
ogy is a niche market. There are a join abruptly at any point in your life the end user. This means the benefit
limited number of manufacturers (e.g., paraplegia after a car accident). of the device needs to outweigh any
out there who make it their business Despite this, accessibility in design is issues of poor aesthetics or stigma
to offer assistive technology. If you often treated as a bonus rather than related to that object. Think for a mo-
need a certain type of device, you a priority. A perceived lack of market, ment about glasses, or as I like to call
have to find one of these companies combined with the unique needs of them, “corrective lenses.” Glasses are
manufacturing an assistive technol- users with disabilities, causes assis- a type assistive technology that has
ogy device to meet your particular tive technology devices to be quite overcome the stigmas associated with
need. Companies can be big or small, expensive. Something as simple as a vision impairment. Fashion designers
they can be local or remote, and, in mount to hold a cell phone or a tablet have hijacked the lens frame and cre-
some instances, they might be the on a wheelchair can cost a $100 in the ated a culture in which a person who
only company that offers the prod-
uct you require. Mainstream compa-
nies don’t always make products with
persons with disabilities in mind, so
Example
even everyday household goods aren’t Designs From
always accessible. (If these compa-
nies were following the principles Our Thingiverse
of universal design, their products
might be more accessible. To learn
Study.
more about a company that does use
universal design, look into OXO.)
For example, a mainstream compa-
ny that makes kitchen utensils might (1) R
 ight angle spoon for people with
offer 13 styles of spoons, but none of limited dexterity; Thing ID: 23729.
them are designed for a person with
low dexterity. Said person with dex-
terity impairment must then seek out
an accessible spoon or an assistive
eating device. Instead of getting four
complete settings of the funky cutlery
with the twisty handles for $35, they
must now buy spoons, knives, and
forks with an augmented handle one
at a time for $10 a piece.
It’s unclear why the assistive tech-
nology market is such a small space. (4) Prototype to convert images to heat (5) P
 rosthetic hand for users
In the U.S. alone, 20 percent of indi- for people with vision impairment; with a functional thumb;
viduals have a disability [1]. Disability Thing ID: 324072. Thing ID: 261462.
is a population that you can be born

72 XRDS • SPRING 2016 • VOL.22 • NO.3

doesn’t even need vision correction
will still pay hundreds of dollars for a
high profile pair of specs. Rapid fabri-
cation offers the ability to revise and
customize an object so that it is not
only a perfect fit for a person’s abil-
ity, it can also be a perfect fit for their
identity and sense of style.
A HANDY EXAMPLE
Let’s jump back for a moment to the
3-D printed hand concept. Perhaps
you own a 3-D printer, which leads ev-
ery member of your family to forward
you every article they find about 3-D
printed chocolates, livers, prosthet-
ics, etc. Or maybe you saw a YouTube
video of Robert Downey Jr. giving a
child a bionic arm similar to the one
worn by his Iron Man character. There
has also been quite a bit of media cov-
erage on groups like e-NABLE (http://
enablingthefuture.org/) and the Open
Prosthetics Project (http://www.open-
prosthetics.org/). Both organizations
have been leveraging the design skills
of engineers and the restless need,
which lurks in the heart of makers ev-
erywhere, to create bespoke prosthet-
ics at reduced costs. It’s inexpensive,
(2) Tactile graphic of Yankee Stadium;
Thing ID: 61329.
(6) Pill bottle lid with daily reminder
label; Thing ID: 84635.
XRDS • SPRING 2016 • VOL.22 • NO.3
highly customizable, and fast.
A child can outgrow a prosthetic
hand every year, maybe even every
six months. The child has to be fit-
ted for the device, any custom com-
ponents need to be manufactured,
and if they’re lucky (by “lucky” I mean
they have access to good, affordable
healthcare coverage) it will be deliv-
ered in a reasonable amount of time
at a reasonable cost. A year later, that
device will need to be exchanged or
at least have swappable components
to size-up with its wearer. With a 3D-
printed prosthetic, you can iterate
through several variations of the same
hand to customize it for less than
$200 and the cost of some dedicated
print time. These 3-D printed hands
are built from tough plastics (think
LEGO) and come in every wild color a
child could want. The prosthetic hand
is no longer a mere assistive technol-
ogy device; instead, it offers a sense
of ownership and empowerment, en-
abling a child to wear their superhero-
heart on their sleeve, literally.
WHAT’S OUT THERE?
As this open-source, printed, pros-
(3) W heelchair mounted environment
controller; Thing ID: 315819.
(7) Ironing guide for individuals
with vision impairments;
Thing ID: 359348.
thetic movement suggests, people are
out there in the world using fabrication
tools to create assistive devices. Online
communities like Thingiverse offer a
place for designers and end users to
develop and share 3-D models. This
open-source site is host to more
than 400,000 designs for 3-D print-
ing and laser cutting, and includes
everything from action figures to
replacements parts for existing de-
vices. In 2014 [3], my lab mates and I
took a closer look at this community
to see what kind of assistive technol-
ogy was already being developed. As
it turns out, there were quite a few
assistive technology designs online.
Back then, Thingiverse was only host
to about 100,000 designs, but it was
still quite an undertaking for us. We
found 363 designs that fit into our
definition of an assistive device.
Here’s what we found:
1. Traditional assistive technology
products currently available as prod-
ucts or made by therapists. Tools for
activities of daily living or math ma-
nipulatives for people with cognitive
disabilities (6 percent of designs).
2. Accessible media. Tactile graph-
ics with Braille that modeled DNA, at-
oms, or buildings (11 percent ).
3. Accessories for assistive devices.
Aesthetic toppers for canes, power
wheelchair joysticks, or game control-
ler joysticks for people with physical
disabilities (8 percent ).
4. Concept designs and prototypes
for assistive technologies. Creative so-
lutions for accessibility challenges or
needs that aren’t currently addressed
by existing products, such as a vision-
ary design of a non-surgical cochlear
implant for the profoundly deaf or se-
verely hard of hearing, or a heat-based
display (4 percent ).
5. Prosthetic limbs. Prosthetic
hands, fingers, or partial fingers for
amputees (17 percent ).
6. Tools for medication manage-
ment. Pill-cutting guides or contain-
ers for sugar cubes for older adults
and people with diabetes (36 percent ).
7. Other designs explicitly intended
for disabled or senior users. Spinner
rings or other sensory fidgets for peo-
ple with ADHD or autism (18 percent).
Most of the designs were for 3-D
printing and not laser cutting, which
73
feature
ACM LEARNING CENTER
RESOURCES
FOR LIFELONG LEARNING
learning.acm.org
Online Courses from Skillsoft
Online Books from Safari, Books24x7,
Morgan Kaufmann and Syngress
Webinars on today’s hottest topics
in computing
74
is likely related to the cost of an entry-
level, consumer 3-D printer versus a
laser cutter. The designs represented
a mix of stand-alone products and
augmentations to existing devices.
These assistive designs also addressed
a wide range of abilities, touching on
visual, motor, and cognitive impair-
ments. So who are the people making
and using these designs?
WHO ARE THE MAKERS?
In a follow up questionnaire sent to
the designers of these assistive tech-
nology items, we uncovered some-
thing very striking—most of the peo-
ple making these assistive devices
didn’t have disabilities. In fact, 76
percent of the designers who replied
to our survey said they did not have
any disability. Several of these design-
ers were making assistive technology
for friends or a family, and there were
also folks designing assistive tech-
nology with no one specific in mind.
The majority of designers were from
technology and engineering disci-
plines and had either gained their 3-D
modeling skills through education,
training, or, most frequently, as a
self-taught hobby. Very few designers
had any clinical knowledge or back-
ground in medical sciences relevant
to the devices they had generated.
This begs larger questions: Why
aren’t people with disabilities lever-
aging 3-D printing for themselves?
What about their caregivers and cli-
nicians, why aren’t they joining in?
How valuable are rapid fabrication
assistive technology products when
A perceived lack
of market, combined
with the unique
needs of users
with disabilities,
causes assistive
technology devices
to be quite
expensive.
XRDS • SPRING 2016 • VOL.22 • NO.3
the end users of those products aren’t
involved in the design?
WHAT’S NEXT?
I don’t have answers to all of these
questions, but I can comment on a
few of them. As I mentioned earlier,
I’ve been working in special educa-
tion trying to find ways to leverage
rapid fabrication tools for accessi-
bility. In so doing, I’ve spent a lot of
time with children and young adults
with disabilities learning to use 3-D
modeling and printing tools, as well
as with occupational therapists and
educators hoping to apply these tools
to existing obstacles. Here are the big
concerns as I see it: 1) awareness and
knowledge of rapid fabrication tools,
2) accessibility of design tools, and 3)
communication and engagement be-
tween communities.
Awareness. A lot of people (perhaps
including you, gentle reader) haven’t
considered the applications of 3-D
printing and other fabrication tools
in the context of assistive technology.
For end users and clinicians, it takes
some examples and a little under-
standing of how 3-D printers or laser
cutters work to start applying those
tools to common accessibility prob-
lems. Once the knowledge is there,
the innovations follow. In some ongo-
ing work I’ve been conducting with
occupational therapists, my lab mates
and I have identified some great uses
for 3-D printing to augment hand-
held tools like pens, pencils, paint-
brushes, and so on for students with
hand dexterity and mobility impair-
ments. We combined low-fidelity pro-
totyping techniques, like modeling in
clay, and transferred those ideas to
digital models to create a robust 3-D
printed solution for a physical acces-
sibility problem. The project has in-
spired more assistive technology de-
sign work and motivated my research
group to pursue new design interac-
tion techniques [4].
Accessibility. Design tools for rapid
fabrication present challenges of ac-
cessibility relevant to novice design-
ers and persons with disabilities. It’s
easy to have an idea for a 3-D printed
object, but it’s harder to generate a
model that will print correctly, meet
expectations, and fulfill metric-
XRDS • SPRING 2016 • VOL.22 • NO.3
If we work to make
fabrication and
design tools more
accessible, we can
empower people
with disabilities
to design assistive
devices for
themselves.
specific demands. Drag-and-drop
interfaces like Autodesk’s Tinkercad
(https://www.tinkercad.com/) enable
the union and subtraction of exist-
ing polygons to create novel designs,
but this method is only accessible for
those with the eye-hand coordination
necessary to design with mouse-driv-
en tools. These beginner tools also
limit a designer’s options as they be-
come more versed and seek to create
more complex designs. The search is
still on for a creative design tool that
bridges the gap between novice and
expert modeler.
Engagement. If you were paying
attention, then you might have no-
ticed that even though Thingiverse
had 100,000 designs in 2014, I only
found 363 unique designs related
to assistive technology. Nearly 400
designs is a great start, but if you
do the math you’ll note that it’s not
even 0.4 percent of what was avail-
able then (and who knows what the
percentage is now). The majority of
those designs were by designers who
didn’t have disabilities, and many
did not work with a person with a dis-
ability when developing their design.
The ability to create customized and
inexpensive assistive devices is a tre-
mendous benefit to persons with dis-
abilities. Yet judging from the distri-
bution of designers in our study, this
group is not completely represented
in the current make-up of 3-D design-
ers and makers. This is not to say the
engineers and technologists lending
their skills to the assistive technol-
ogy cause are unwelcome. Instead, I
would ask, how do we diversify these
designers and create more opportu-
nities for designers with disabilities
to self-design and become further
incorporated into the making move-
ment?
CONCLUSION AND A CALL
TO ARMS (OR HANDS)
Fabrication tools offer a lot of great
ways to support and create accessibil-
ity addressing issues of customization
and cost. For you starry-eyed current
and future researchers and technolo-
gists, consider joining in the fun. Ac-
tionable steps that we can take to sup-
port accessibility by way of fabrication
might include simplifying tools; cre-
ating alternative design interfaces;
or providing new tutorial systems to
help educate end users, clinicians,
and future designers. We should con-
sider the abilities and goals of these
potential assistive technology mak-
ers (and users) and work with them to
uncover best practices for fabrication
tool design. Ultimately, we can gen-
erate greater access for persons with
disabilities, develop more opportuni-
ties for assistive technology design,
and create a more diverse community
of makers and researchers.
References
[1] U.S. Census Bureau Public Information Office. Nearly
1 in 5 people have a disability in the U.S. Census
Bureau Reports. U.S. Census Bureau. Press release.
2012. https://www.census.gov/newsroom/releases/
archives/miscellaneous/cb12-134.html.
[2] Phillips, B. and Zhao, H. Predictors of Assistive
technology abandonment. Assistive Technology 5, 1
(1993), 36–45.
[3] Buehler, E., Branham, S., Ali, A., et al. Sharing is
caring: Assistive technology designs on Thingiverse.
In Proceedings of the 33rd Annual ACM Conference
on Human Factors in Computing Systems (CHI ’15.)
ACM, New York, 2015, 525–534.
[4] Buehler, E., Kane, S.K., and Hurst, A. ABC and 3D:
Opportunities and obstacles to 3D printing in special
education environments. In Proceedings of the
16th International ACM SIGACCESS Conference on
Computers and Accessibility (ASSETS ’14). ACM, New
York, 2014.
Biography
Erin Buehler is a fifth-year Ph.D. student in the Human-
Centered Computing program at the University of
Maryland, Baltimore County advised by Dr. Amy Hurst.
Her research supports universal access to education for
students with intellectual and developmental disabilities.
Buehler’s work has explored the use of rapid fabrication
tools and individualized interface design to improve the
accessibility of curriculum in both formal and informal
educational settings.
© 2016 ACM 1528-4972/16/03 $15.00
75
feature
PROFILE   DEPARTMENT EDITOR, ADRIAN SCOICĂ
Dennis Bormann
The Man Who Introduced Antarctica’s
Davis Station to 3-D Printing
DOI: 10.1145/2893519
Dennis Bormann lives
in Australia, and has
been passionate about
mechanical engineering
for the better part of
his life. Having left
high school early to start a career in
the maintenance and improvement of
mechanical equipment, his incrementally
accrued experience ultimately made
him an ideal candidate for working with
the Australian Antarctic Division at
the Davis Station. It was there, on the
barren shores of Antarctica, Bormann
inadvertently ended up sharing his
hobbyist knowledge of 3D-printing
with his scientist colleagues, thus
demonstrating engineering applications
and possibilities they had no previous
first-hand acquaintance with.
Upon completing his 12-month long
Antarctic assignment, Bormann returned
to Australia to continue pursuing his
formal education. He agreed to an
interview with XRDS to share his personal
story of exporting 3-D printing to remote
places inhabited by penguins.
ALWAYS A TINKERER
Bormann’s story of tinkering with
machines started at the age of 13. This
was when he took an interest in off-road
vehicle-building; a hobby many would
find unusual for his age. Even though
he started out under the auspices of
older friends who supported him in his
endeavors, he slowly transitioned to
working independently by the age of
16. In 1989, he left high school upon
the completion of year 11 to start an
apprenticeship with William Adams
(a local dealer for the internationally-
renowned machinery corporation
Caterpillar).
“The apprenticeship with Caterpillar
was four years just on the tools, and after
76
that I spent another five years with the
company, bringing the total to nine years
between 1990 and 1999,” confessed
Bormann. Duirng that time he acquired
knowledge about the mechanics of a wide
range of earth moving gear—excavators,
loaders, bulldozers, and scooters—which
would later set the scene for his next
career move.
At the end of his time with Caterpillar,
Bormann decided to broaden his
experience. In 1989 he went to work for
a local aircraft company called Gippsland
Aeronautics (today known as GippsAero,
following a 2009 acquisition by the
Mahindra Group Indian conglomerate).
“I was doing a lot on the production line
for them. It was initially maintenance
and customer support, and from
there on I moved to their Research
and Development division, looking
for ways that they could improve the
processes that we were going through,”
remembered Bormann. The job allowed
him to get around a fair bit, with regular
trips to the Mojave Desert in California,
and occasionally to places like Hawaii
and Canada.
Bormann had his first contact
with 3-D printing at GippsAero, during
the technology’s early days circa
2000–2001. Even though GippsAero
turned out to be uninterested in
incorporating 3-D printing into their own
technological processes at the time,
Bormann followed the development of
3-D printing for another 10 years. By
2010 it became apparent the field was
transitioning from hobbyist status to
mainstream consumer status.
After leaving GippsAero in 2009 ,
Bormann went to work as a mechanic
again, but this time in the ski industry.
While his main job was to look after snow
groomers, ski lifts, and snowmobiles at
the smallest of the three main ski resorts
in his home state of Victoria, he still
kept an eye on the developments in 3-D
printing as a personal hobby.
It wasn’t until 2013, upon noticing the
3-D printing field had become sufficiently
mature that Bormann finally decided to
purchase his very own 3-D printer—an
experimental one released by MakerBot.
“I thought the time was up... 3-D printing
was finally at that point where I could put
money in it, and off I went with it. And it
was one of those things, where seeing
what everyone else was doing, I knew that
as soon as the patents expired, there was
going to be a boom—which there surely
was. It’s only been growing from there
on,” he explained. Borman picked the
perfect time to invest in his hobby, and
has since never looked back.
APPLYING FOR A JOB
IN ANTARCTICA
Having capitalized on his prior
experience, Bormann proceeded to
make the most inspiring, and perhaps
most fascinating, move in his career
to date: Applying to work with the
Australian Antarctic Division, despite his
uncertainty about getting accepted.
“Applying to work there was
something that I had always wanted to
do, but at the same time I always thought
that I’d never get a job there. I just put in
the application,” he recalled, adding he
had cautiously managed his expectations
throughout the process.
However, it turned out, he found
himself in a position where he had the
majority of the bases covered for the
position as a mechanical supervisor. The
application process was very rigorous,
and all in all it took a little over six
months, during which he underwent very
stringent medical exams in addition to
mechanical aptitude tests. He emerged
victorious. He was offered a position as
XRDS • SPRING 2016 • VOL.22 • NO.3
 a station mechanical supervisor (SMS)
in November 2014. His role would be to
oversee everything mechanical—power
generation, fuel use, maintenance,
and servicing all the client equipment
generators that back up the entire
scientific program.
LIFE AT DAVIS STATION
Talking about what life was like at
Davis Station, Bormann is quick to
dismiss the common preconceived
image of cramped and uncomfortable
quarters hanging off the edge
of civilization. “ The best way to
describe it is if you’ve ever worked in
a mining camp, or on an oil rig. In the
sense that there’s a very close knit
community, and only a certain number
of faces that you see every day. You
create tight bonds with the other
people on the station, and you all work
together. But otherwise, conditions
are pretty much the same that you
have at home. There’s a small cinema,
a bar, a pool table, and entertainment
to keep your mind off of being so far
out. As for the work side, besides the
harsh winds and the colder climate,
they are not that much different from
the mainland,” he said, adding there
are also aspects that make working in
Antarctica unique.
“One thing is you really need to think
about how you’re doing your job. If you
get yourself injured, there’s only one
doctor on station, and you can’t get
yourself evacuated, especially in winter,
so you can’t take unnecessary risks.
Also, another thing is you can’t just pick
up and go to the corner shop for bits
and pieces to make repairs. You have to
come up with solutions using the parts on
the shelf. These, I think, are the hardest
adjustments, besides leaving your family
at home.”
Bormann emphasized the hardest
adjustment he had to make was
witnessing the changing of the
seasons. When the austral summer
ended around March, and temperatures
fell, a ship came to pick up a portion of
the team, dropping the population on
the base from 90 to 21. Come winter,
the wildlife around the base—mainly
Adelie penguins and various species
XRDS • SPRING 2016 • VOL.22 • NO.3
of seals—also disappeared, leaving
behind a desolate and dark landscape
made harsher by the knowledge that
no ship would be coming back for quite
a while. “When all of the summer crew
left, and the ship was sailing away, it
was quite a surreal feeling that hung
on: that those left of us were going to
be here for the next six months without
anyone else,” Bormann recalled.
To manage the sense of isolation,
people could resort to the Internet or
phone line to call home every other day.
But more importantly they were advised
to dedicate their extra time to developing
their hobbies.
TAKING 3-D PRINTING
TO ANTARCTICA
According to Bormann, station residents
are encouraged to ship whatever
they need to the station, so they can
continue with their hobbies when they
had down time.
He already had a MakerBot printer
at home, so he decided to buy a new
PrintrBot kit and had it shipped it
with the rest of his belongings. After
he initially spent a lot of his spare
time improving his skills at drawing
3-D models with open-source
software, he sparked the curiosity
of his fellow station members, two
of whom approached him with little
side projects they had been working
on themselves. His colleagues were
curious whether components of their
own projects could be printed, and
before long Bormann ended up
making quite a few things for different
people, including a replacement part
for a broken tripod, stamps for
leather working, and numerous
Christmas gifts.
His hobby caught on.
After this initial experience, other
colleagues, who had witnessed what
could be achieved, ended up ordering
printers for themselves as well. From a
scientific point of view, many remarked
on the future benefit of having a 3-D
printer on site. “They’d come to me
saying they had an idea, and I would just
draw it out, and by the next day I’d hand
them a piece, and they’d be amazed.
I ended up showing more people—and
they were a captive audience—just
what you can do with a 3-D printer down
there,” he happily reminisced.
BACK ON THE MAINLAND
Now back in Australia, Bormann
is focusing his energies toward
completing his bachelor of engineering,
while he continues to use 3-D printing
as a tool to prototype his ideas and
concepts. He is very grateful for the
invigorating experience of having
worked with some of his country’s
brightest people. He would strongly
recommend his Antarctic experience
to others wishing to contribute toward
something bigger:
“One thing, especially about
Antarctica, is that you don’t have to
be a scientist to go there and make
a difference. There are a lot of roles
that need to be filled, whether as a
medical staff, being a chef, looking
after the station’s water supply, being
a carpenter, or being a computer
engineer taking care of ICT support and
communications. Even as a tradesman,
you could be sitting down and having
dinner with one of Australia’s leading
scientists,” he said. Bormann’s advice
to young people is to aim high and
believe in themselves regardless of
their background. Who knows where
your hobby may lead you.
Copyright held by Owner(s)/Author(s).
77
end
Dr. Kihong Ku and students discuss a project prototype fabricated by the CNC router.
LABZ
Design Informatics Lab
Philadelphia, Pennsylvania
T
he Design Informatics Lab
(di-Lab) explores emerging
digital methods and material
technologies for architectural
design. di-Lab is housed within the
College of Architecture and the Built
Environment (CABE) at Philadelphia
University. Since 2011, the lab has
been used by graduate and under-
graduate students with backgrounds
in architecture, electrical, mechani-
cal, textile engineering, and other de-
sign disciplines. Students interested
in digital methods work with various
geometric modeling programs, visual
programming languages, and rapid
prototyping tools. Students investi-
gate and develop design solutions
for buildings and/or building sys-
tems (i.e., envelopes). The three major
themes of the lab are 1) design and in-
78
formation modeling, 2) intelligent ar-
chitectural systems, and 3) emerging
technologies and materials; each has
a different logic and enhances design
in various ways. The goal is to prepare
students to help further the field of ar-
chitecture by introducing these evolv-
ing design methods.
The architecture school teaches
the fundamentals of creative design,
which often exclude advanced com-
putational design processes. In the
di-Lab, students focusing on emerg-
ing technologies and materials are en-
couraged to use various digital meth-
ods—from early schematic design all
the way to prototyping. Students have
the opportunity to explore and dis-
cover interdisciplinary architectural
design in different ways beyond using
CAD and rendering software.
3-D modeling software is an essen-
tial tool used to represent designs. In
the lab, besides learning Rhino 3D (a
geometric modeling software), stu-
dents learn how to use open-source
software such as Grasshopper. It is a
customizable, visual scripting plugin
that enables architects and design-
ers to define parametric relationships
and to program without being fluent
in computer languages like C#, Visual
Basics, or Python. Visual scripting al-
lows users to create individual defini-
tions, which help to generate complex
geometry for architectural designs or
systems and to incorporate charac-
teristics of physical materials or fab-
rication techniques into geometric
constructs. Adopting various design
tools enables students to complete
tasks requiring physical simulations,
which can sometimes be difficult or
even impossible to do manually. Thus,
students working in the lab build
their designs through research that
advances their findings using digital
simulations and material iterations.
Our own research experience has
brought a variety of subjects into ev-
ery project, requiring us to think criti-
cally at all times.
With innovative architectural de-
sign being the main focus, students
in the lab conduct extensive research
on highly specific topics. Biological fi-
ber and textile engineering is one such
emerging topic. In a fiber structure
project, one of us (Billy Wong) closely
studied the basal layer of human skin
and discovered its structural signifi-
cance and filtration function. Using
the findings, he was able to represent
the structure as a 3-D model. The re-
sulting envelope design was proposed
as a self-sustaining structure and also
a parametric response to natural light-
ing and ventilation.
In another hands-on project, we
(Javier Villarroel and Feras Alsaggaf)
developed non-uniform, stackable
fabrication modules, which allow sim-
XRDS • SPRING 2016 • VOL.22 • NO.3
 $5B
The expected global market share of 3-D printing
by 2018, analysts also predict the technology will
become 50% cheaper and up to 400% faster.

ple wall partitions to be built at ease. BACK

In order to bring a 3-D model to proto-
typing, data was first translated from
visual scripts to G-Code and M-Code.
(This is a list of instructions for a CNC
router, which is a milling device that
fabricates models by subtracting ma-
terial.) Using a CNC router, we ran
numerous test trials on various spe-
cies of wood at different thicknesses.
After three weeks of trial and error,
the team was able to fabricate novel
joints, textures, and forms. All routed
modules were immediately ready to
be assembled.
The lab endeavors to push the
boundary of architectural innovation
using digital methods and emerg-
ing materials. Projects coming out of
di-Lab include architectural robot-
ics and novel architectural compos-
ites for building skins. Students have
also actively participated in a num-
ber of international design competi-
Stratasys FDM 2000 image by Chris Davis (Flickr). Stratasys uPrint SE Plus Image by Biswarup Ganguly (Wikimedia).
The Emergence
of 3-D Printing
Prior to 1952, fabrication had always been analog until a group of
researchers at MIT were able to connect a milling machine to a
computer. The computer used a punched tape to feed instructions
to the machine, thus making it the first experimental numerically-
controlled machine—opening the doors to the “digital” fabrication
era. Early digital fabrication was, however, limited to “subtractive
manufacturing,” which involves the removal of excess material
until the desired shape is reached. Hence, a product with an inner
structure was created incrementally by fabricating its composing
solid components, which were then fed into another process that
assembled them into the final product.
Then, in 1980, 3-D printing came to life. A product could be built in
a single run; a designer would feed the design to the printer and the
finished product would come out in one piece. This process is called
“additive manufacturing.” Currently, 3-D printers are slow, limited
to relatively small build sizes, and not yet suitable for manufacturing
large whole products. For this reason companies rely on 3-D printers

tions. Based on the lab’s work, CABE to produce early product prototypes in a process called “rapid
was one of three academic programs prototyping.” But that isn’t stopping today’s kids from designing their
in the U.S. that received the 2015 Na- own toys and having them fabricated at their nearest Makerspace.
tional Council of Architectural Reg- Below is a comparison between two Stratasys 3-D printers that use
istration Boards Award. The program fused deposition modeling (FDM) technology.
won $34,208 to develop strategies for —Asmaa Rabie
architectural textile composites for
building envelopes. This project will
foster interdisciplinary collaboration
among architecture, textile, fashion,
and industrial design and engineering
students on projects that cover a broad
range of subjects.

Biography
Feras Alsaggaf is a fifth-year architecture student at
Philadelphia University. He is devoted to learning emergent
technologies in visualization and prototyping and their
application to architecture so materials and structural
assemblies are tested and pushed to their limits.
Javier Villarroel is an undergraduate architecture student
at Philadelphia University. He is often fascinated by FDM 2000 uPrint SE Plus
biological systems. Villarroel aspires to be involved in the
development of unconventional building materials with the Production Year 1997 2011
goal for optimizing the built environment while enhancing Layer Thickness 0.007 ~ 0.014 in 0.01 ~ 0.013 in
the natural habitat.
Build Size 254 x 254 x 254 mm 203 x 203 x 152 mm
Billy Wong is a recent graduate from the architecture
program of Philadelphia University. He is a versatile Material Colors ABS(White) ABSplus (9 colors)
learner who adapts to all kinds of software as design tools. Price $120,000 to $140,000 $29,900
Wong is currently developing real-time architectural
visualizations by utilizing virtual reality technology and 660 x 914 x 1076 mm 635 x 660 x 787 mm
Size & Weight
game development engine. 160 kg 76 kg

XRDS • SPRING 2016 • VOL.22 • NO.3 79

Check out the new acmqueue app
FREE TO ACM MEMBERS
acmqueue is ACM’s magazine by and for practitioners,
bridging the gap between academics and practitioners
of computer science. After more than a decade of
providing unique perspectives on how current and
emerging technologies are being applied in the field,
the new acmqueue has evolved into an interactive,
socially networked, electronic magazine.

Broaden your knowledge with technical articles

focusing on today’s problems affecting CS in
practice, video interviews, roundtables, case studies,
and lively columns.

Keep up with this fast-paced world

on the go. Download the mobile app.

Desktop digital edition also available at queue.acm.org.

Bimonthly issues free to ACM Professional Members.
Annual subscription $19.99 for nonmembers.
 HELLO
HELLO WORLD
WORLD

Headline here Run-in dek

The Brownian
follows hereof
Wanderlust lorem ips
Things
U
giatent, eate diatiae nos emporemodis de reprat. dentem esenimu sanimus volorporis
BY MARINKA ZITNIK
dicid quissed quate et res Ucid et volorit que venda num nonsend denis as volut omnis ab idessimusam

C
et archit qui illibus reriorro icipsaest maxim ute sit eat ipsandisto arit, cupta acimilit quodis voluptatque
mintiusarecto
onsider gambler exeratet officti
who starts modissit abora
the volatility commoll
refers to the uptioss itatur
influence of pra
to velesunderstand
better simintibusahow iuntideas
re volorem
repedis modistio eosamfortune
with an initial assuntand etur, in aut vere, officiaevents
unpredictable cullatur in sinvelis
our model. ut acimini atendani
nis expel disseminate in onlineiduciendeles
networks asnes ea
nient aruptus playseicate ea que velit
the following simpleofficias
coin eumThe assum etur? of n steps of
calculation si doloresto
well dolorum eiumqui
as via non-Twitter and non- beatect
ipsum nimus tossingpostem game. nost,At eachnis modistis
turn, Ur, nonse
Brownian sam facimi,
motion is actuallyipsumverysoles di
simple. otatust explitatus
Facebook channels,aut suchquati cus del
as word of
culpa
the dolum
dealer que laut
throws auditiundicoin.
an unbiased rehendi quid quodist
In Listing 1, weiisquis
firstnon nihilicid
generate allmagnam
the modi doluptam
mouth and newsapedita
media. The de nis
wayeosin
tatem.
If Udaest, is
the outcome siminventiam
head, the gambler sum doluptatem
samples from iumthe re normal
od endisima core
distribution assequi
which stiunte moditas
information et moluptati
flows within the
ut peres
wins a unit;quoifet thevellaut adi cusup
coin comes autes
tails, magnis
with oneelcallinum toreperibus quuntur?
scipy.stats.norm.rvs() veleste
online eatem accum
networks affectseamass
nonectotaqui
media
est pa
the perumqu
gambler loses odiscipsape
a unit. Theoptiur gambler andLorit
thenautform raetheporporepudi
cumulative con
sumnima optatiori
and offlineconescilit,
spread of ut optae il ero
information [5].
reptis postioratur
continues to play until rerchilhe ibuscium
is either laut audae
using thenamNumPy
ad ulparchil
cumsum mosanih illatem
function. beaque nitas
However, to bearcius
able toeatem quuntiunt
truly capture
omnima postis
bankrupted or hisdoluptatiae.
current holdings Idelibea ereptiam aut apides
Figure 1 (left) expla plot
is a typical consequam,
showing omnihilignisinrestrumquia
propagation the Twitter nimus des
environment,
volorepsome
reach tatatio.fixed Etdesired
omnit praturiamount. If cones
several eaquiatem
realizations estof peBrownian
pra doluptatmotion nonestin
we need toeture
take aborehe nimporro
into consideration
onsequis
each coinescium non re et rempos
toss is independent, then we maionsed
generatedquis by theet hillorrunte
function defined quia quodi
in doluptatque
the stochasticventibus
nature of consequid ut uta
Twitter topic
estrum
say ium estruntibus.
the balance of the gambler has a cullam,
Listing uteneste
1. The same enisfunction
a quiamcan coreste venist optur That
be used propagation. aut earciurio.
is, we wouldEt unt
likeidel
to
Ulparibusam
distribution of a volorporati
simple random reiunt,walk. modignimo
to generateipsum Brownian ex eamotion
dus esinqui volupti model
higher istionsed et, imet que
the intrinsic consendantus
random nature of
nissimo lorepro
In addition toodipien
the fortune itatur? of a con et faccabsince
dimensions, incieteachidisidimension
bernatis nobitiis quatibeatust
is just individuals volupta cus.
to potentially adopt and
Test expedi
gambler flippingdis es autclassical
a coin, facipienis et magnihit faciur, consenit,
a one-dimensional Brownianod et quist
motion. Ratus escia
eventually voloratis
broadcast archilles
a Twitter re
topic.
alias et plam,
examples volo quae
of random walks natqui aut
include eum eatia dolleca
For example, Figure borepratquis
1 (right) showseos et a erum
One into corromodel
stochastic oditatur,
thattem fugitin
is well suited
exceruptur, ut
fluctuations ofet thelitstock
eos est marketetur [1] etur?
typical trajectory of a two-dimensional ped
to ute
this proisqui
task omnisqui
Brownian coria debisin
motion [6].
autem
and ipsam, motion
Brownian ate nosapel [2], whichet quiatur,
we Voloria motion
Brownian sitatquiwhose ute quiasistartqui
andnonsequi
end recum di sed qui sam dia nus eatia
ipidempore
examine sequate
in this column. mquaspellore
We describe a bla cusae
points aresunt unt maximinveles
represented with green ipsandit
and BROWNIAN
pratem voluptateDISTANCE
nullupt atempos
quibus etparadigm,
popular assimus voloruntio.
which translates Ut peruma aligenem
red markers, alibeature as namWe
respectively. rest occab
next AND TRUST
reprovit FUNCTIONS
voleni omnis pel ipsandisquis
qui sit, que pre, qui
one-dimensional a consectatet
stochastic process unt illorum
describe apehow disque
Brownianre es volupta
motion cantquundebe doluptam
Suppose wequibus
represent Twitter activity
faccus eatibus
illustrated above pato doluptatiis
networks autatem (graphs). ndiciis
used tocum, model id quo omnisci aturist,
information propagationsero with a toy network of mentions shown
essequam
In the context alitatibusa
of networks, volor aalic tent
random te
overnones experibus,
social networks. ipsam qui vel eost et in Figure 2. Here, each node represents
faccus.
walk typically corresponds to a process etur? Lic testrunt eos deni ant. a Twitter user. There is a directed edge
Tiorem
in which aliae. Nem
a walker moves harum from one node THE OrumSPREAD OF INFORMATION
as velliatium ant que ese prae from user i to user j if i mentions j in
assimagnat
of the network labintoipitamentio
another with tem fugit
a lit siINulluptatem
SOCIAL MEDIA ut rem OUTLETS
sundis sin cus ex a tweet, and the edge label denotes
aboriti stotatur,
probability quistiantiur
proportional to the sitia dolo of
weight et eatempelit,
Modeling optaquiofcorerep
the spread eriandi
information the number of tweets in which user i
quiaepr
the edgeenditio.
connecting Nus earum
them [3, esserferro
4]. Let’s consequi
within socialtet harchicatur?
media outlets, such as mentions user j. It is expected [5] that
dolesci demperu ntotate mperuptur
start! Axima
Twitter pratinctati
and Facebook, omnia as acipsam
is important central players, such as celebrities
mi, ent. dolorestem. Berspis quiscil lacium
BROWNIAN
Offic tem. Ovid WANDERER que conestor esequi
Listingduntem rehendandae
1: A Python function natus
thateximplements one-dimensional Brownian
sin estrum
Random walkquiisdusam que nobitib
a mathematical earitioria
motion.dolorGivenmolor maximi,
the initial accatiisX , the number of steps to take n, the time
condition 0
eaquaestiur rest
formalization of arepathdemresulting
fuga. Quifrom an aut audictis
step dt , and dolut
thelist ut iliquiatus
volatility quibusc
parameter δ, the script generates an instance of
dolorem
object oluptatur
taking successiveaborrorandom beatemsteps imuscipsam rerione peles ea vollabore
Brownian motion.
volupta
in a domainsumspace.explace Westemquo
considermin here a raeperc hilitassi volum harchil luptas et
prehendi nempore,
particular con pligent
type of stochastic repro
processes dolorumfrom aliquibus.
math import sqrt
odia nam
known as sapieniania
Brownian motion. quasitiOne blaborum
form of Abor antiam laccuptas erferferfero
from scipy.stats import norm
que model
the coribearum,for Brownianvitenimod motion quasisutgiven eum nimpedi ipiet hiciis quidicipsant
import numpy as np
ditatis
by sed quiderum
the initial condition, aut i.e.autthelastarting
sequi la inventu riost, inis erunt lab inus,
con consendem
position of Brownian cus. motion, X(0)= X 0 , incius essum nimi, tet ad magniendi
def brownian(x0, n, dt, delta):
andRibus iduciat velluptat
the iteration rule, X(t+dt)= lacepra X(t)+ coreptaecaex0que = ius
np.asarray(x0)
aut omnis mod quia
r = norm.rvs(size=x0.shape + (n,), scale=delta*sqrt(dt))
sandunt
N(0,δ 2
dt), aribus
whereque diam
N(0,δ 2
faccumquo
dt) is a normally expe nimus et mos im erspe etur alitatem
brown = np.cumsum(r, axis=-1)
to doles cullab
distributed randomiducilitaqui
variable volestium
with mean vellit de remperumbrown += velnp.expand_dims(x0,
ipsandem inum, axis=-1)
disi
0 andcone ommolo
variance duciet
δ2 dt. earum illibus
Per convention, eaquodis idenit returnliaspisquas
brown aut volupta
concall
we nulliati
these offictur, volupis doluptam
two parameters drift and tiatibus.
>>> x = brownian(x0=10*np.ones(20), n=500, dt=10./500, delta=2)
is atur? respectively [1]. The drift
volatility, Lorporepre dollore pudit, ulles endam
Non necaboribus
represents deterministic dolorer ibuscit
trends, while ipient lignatemodis accum enihil et modis

XRDS • SPRING 2016 • VOL.22 • NO.3 81

 HELLO WORLD

Headline here Run-in dek

Figure 1: Left. A typical plot showing 20 realizations of one-dimensional Brownian motion. For each realization,
500 steps of Brownian motion were generated using the script in Listing 1 and setting the initial condition to X 0=10.
Right. A typical plot showing Brownian motion in two dimensions.

follows here lorem ips

U
20 –0.0

giatent, eate diatiae nos emporemodis de reprat. –0.1 dentem esenimu sanimus volorporis
15 dicid quissed quate et res Ucid et volorit que venda num nonsend denis as volut omnis ab idessimusam
et archit qui illibus reriorro icipsaest maxim ute sit–0.2 eat ipsandisto arit, cupta acimilit quodis voluptatque
mintius recto exeratet officti modissit abora commoll uptioss itatur
–0.3 pra veles simintibusa iunt re volorem
10
repedis modistio eosam assunt etur, in aut vere, officia cullatur sinvelis nis expel ut acimini atendani iduciendeles nes ea
–0.4
nient aruptus eicate ea que velit officias eum assum etur? si doloresto dolorum eiumqui beatect
X

Y
ipsum 5 nimus postem nost, nis modistis Ur, nonse sam facimi, ipsum
–0.5 soles di otatust explitatus aut quati cus del
culpa dolum que laut auditiundi rehendi quid quodist iisquis non nihilicid magnam modi doluptam apedita de nis eos
–0.6
tatem.0 Udaest, siminventiam sum
doluptatem ium re od endisima core assequi stiunte moditas et moluptati
ut peres quo et vellaut adi cus autes magnis el inum reperibus –0.7quuntur? veleste eatem accum ea nonectotaqui
est pa perumqu odiscipsape optiur Lorit aut rae porporepudi con nima optatiori conescilit, ut optae il ero
–5 –0.8
reptis0postioratur 2 rerchil ibuscium
4 laut6 audae 8nam ad ulparchil 10 mosanih
–0.1
illatem
0.0 0.1
beaque
0.2 0.3
nitas0.4
arcius
0.5
eatem
0.6
quuntiunt
0.7 0.8
omnima postis doluptatiae. Idelibea Time ereptiam aut apides expla consequam, omnihilignis Xrestrumquia nimus des
volorep tatatio. Et omnit praturi cones eaquiatem est pe pra doluptat nonestin eture aborehe nimporro
onsequis escium non re et rempos maionsed quis et hillorrunte quia quodi doluptatque ventibus consequid ut uta
estrum ium estruntibus.
and politicians, are more likely to be cullam, utenestethe
until it reaches enis a quiam
trust coreste
threshold after BROWNIAN
venist optur aut PROPAGATION
earciurio. Et unt idel
Ulparibusam
mentioned by other volorporati
users than reiunt,
the other modignimo
which new information dus es quiavolupti STRATEGY
ipsum ex eaincreases istionsed et, imet que consendantus
nissimo
way around. loreproNotice odipien
in theitatur?
network from con et faccab
user’s inciet idisi
trust linearly. bernatisinnobitiis
We assume our quatibeatust
Given the mentions volupta cus. in Figure
network
Test2expedi
Figure user 4 dis es aut facipienis
mentions user 3 eight et magnihitnetwork
mentions faciur, consenit, od et
a user’s trust in quist
a topic 2, whichRatusisescia voloratisdistance
the Brownian archilles re
alias
timesetwhile
plam, uservolo3 quae
does natqui
not mentionaut eum eatia
crosses dolleca
the borepratquis
threshold when he or eosshe et erum into
matrix corro oditatur,
calculated from thetem fugitin
network
exceruptur,
user 4 at all ut (the et reciprocal
lit eos est edge
etur etur?
has tweeted about it. Modeling the trust ped ute pro qui
connectivity andomnisqui
the trustcoria debisin
functions, we
autem
with zero ipsam,
weight ateisnosapel
not shown et quiatur,
in the Voloria with
functions sitatqui ute quiasi
Brownian motionqui allows
nonsequi now recum di sed the
describe qui algorithm
sam dia nus foreatia
Brownian
ipidempore
network). sequate mquaspellore bla cusae
us to sunt unt
naturally maximinveles
account for stochasticity, pratempropagation
ipsandit motion voluptate nullupt atempos
[5]. Suppose a
quibus et assimus
Following the work voloruntio.
in Jin etUt al.perum
[5], aligenem
which might alibeature
arise due astonam rest occab
information reprovit
user i postsvoleni omnis pel ipsandisquis
an infection-related tweet
qui sit, quethe
we define pre, qui a consectatet
Brownian distance unt illorum
diffusion ape disqueofrethe
outside es Twitter
voluptaortquunde doluptam
at quibusindicates i has been
time t 0 , which
faccus
between eatibus
two Twitterpa doluptatiis
users to autatem
be d ij =1/ ndiciis cum, of
occurrence id unpredictable
quo omnisci aturist,
events.sero recruited or infected. In Figure 2, node
essequam alitatibusa
((ωij +1) (ωji +1) γ
), wherevolorωij is alic
the tent
weight te nones experibus, ipsam qui vel eost et
faccus.
of directed edge from i to j. Here, we use etur?
FigureLic 2:
testrunt
Simulation eos deni ant.
of Brownian motion propagation algorithm [5] on a toy
theTiorem aliae. Nem
bias coefficient harum
γ≥0 to give more Orum as velliatium
Twitter mentions network with ant que ese prae eight users. Green node denotes a user who
assimagnat
weight to mentions,lab ipitamentio
which carrytem fugit
morelit si posts
ulluptatem ut rem sundistweet
a protest-related sin cusatex time Time=0, while red nodes indicate users
aboriti
impact.stotatur, quistiantiur
If user i mentions usersitia dolo
j, that etwho
eatempelit, optaqui corerep
were recruited later in time. eriandi Edges are weighted by the number of tweets
quiaepr enditio.
is ωij >0, this Nus earum
expresses esserferro
i’s intention consequi
in whichtet harchicatur?
source users mentioned target users.
dolesci demperu
to propagate ntotate mperuptur
information to user j Axima pratinctati omnia as acipsam
mi,
[5].ent.
However, j might ignore i and may dolorestem. Berspis quiscil lacium
γ=2
Offic tem. Ovid
consequently rarely que conestor
mention user i. We esequi duntem rehendandae Time = 1 natus ex µ=1 Time = 3
sin
thusestrum
up-weight qui dusam
the returnque nobitib
mentions earitioria dolor2
molor
3 maximi, accatiis σ = 1.5
2 3
eaquaestiur
ωji by γ. In ourrest re dem fuga.
experiments Qui toy
on the aut audictis
1 dolut list 2 ut iliquiatus 7
quibusc 1 2 7
5 1 5 1
dolorem
network we oluptatur
set γ=2. aborro beatem imuscipsam 8 rerione peles ea vollabore 8
6 1 4 6 1 4
volupta sum explace
We proceed stemquo
by describing min
trust raeperc hilitassi
4 volum2harchil 2 luptas 1 et8 4 2 2 1
8
prehendi
functionsnempore,
[5], whichcon are pligent
used torepromodel dolorum aliquibus. 1 6
3 1 6 3
4 3 4 3
odia nam sapieniania
an individual’s agreement quasiti
withblaborum
an idea Abor antiam laccuptas erferferfero
3 5 3 5
que
thatcoribearum,
is expressedvitenimodin tweets. quas The trustut eum nimpedi ipiet hiciis quidicipsant
Time = 6 Time = 9
ditatis
function sedis aquiderum
function aut of twoautusers
la sequi la inventu riost, inis erunt lab inus,
3 3
con consendem
between whom trust cus. is modeled. It is a incius essum 2 nimi, tet ad magniendi 2
1 2 1 2
Ribus iduciat
Brownian motionvelluptat
stochastic lacepra
process coreptaecae que
5 ius aut omnis mod
7
1
quia 5
7
1
expe 6nimus 8 et mos im erspe etur alitatem 8
sandunt aribusvalues
with particular que diam faccumquo
for drift and 1 4 6 1 4
4 2 2 8 4 2 2 8
to doles cullab
volatility that can iducilitaqui
be estimated volestium
using a vellit de remperum vel ipsandem 1
inum, 1
3 3
disi cone ommolo
maximum likelihood duciet earum
approach illibus
[5]. The eaquodis 4 idenit
3 liaspisquas1 aut 6volupta 4 3
1 6
con nulliati
intuition offictur,
is the volupis
following: doluptam
Initially, when tiatibus. 3 5 3 5
is atur?
a user receives new information, that Lorporepre dollore pudit, ulles endam
Nontrust
user’s necaboribus
increasesdolorer ibuscit
exponentially ipient lignatemodis accum enihil et modis

82 XRDS • SPRING 2016 • VOL.22 • NO.3

 5 corresponds to such an initially
infected user. User i will then infect
its neighbors based on the values of
its neighbors’ trust functions with
user i. For example, if j is a close
friend of i, then it is more likely that
j will be infected relatively quickly
because of j ’s trust in i. However,
if user j is only a distant friend of i,
then it might take a long time before
j builds trust with i and accepts i ’s
status. The model in [5] assumes
that only after j ’s trust with i reaches
some threshold, user j agrees to
adopt i ’s opinion. We consider the
Brownian distance d ij to be the
trust threshold. Since the trust
functions grow continuously over
time, given enough time for diffusion,
all neighbors of an infected user will
eventually get infected.
Figure 2 shows a typical realization
of a propagation through our
network. In the figure, the green node
represents a user infected at the
start of simulation and red nodes
denote users that get infected later in
the process. Since Brownian motion
is a time-continuous stochastic
process, we discretize the time using
steps of fixed duration. We show our
toy mentions network and currently
infected nodes at four points of the
simulation. (The Brownian motion
propagation algorithm is detailed
in Jin et al. [5] and our Python
implementation of the algorithm is
available as the online supplementary
material to this column.)
MODELING MASS PROTEST
ADOPTION ON TWITTER
The Brownian propagation strategy
described above was developed
by Jin et al. [5]. They applied their
strategy to the Twitter mentions
networks related to 40 different
social unrest events that occurred in
Latin America since 2012. Different
protests were identified by unique
Twitter hashtags. For example,
the education reform protests
in 2013 in Mexico were collected
based on the #ReformaEducative
hashtag, whereas the social
protests against violence and
crime in Uruguay in 2012 were
XRDS • SPRING 2016 • VOL.22 • NO.3
identified by #UruguayosIndignados
and #HartosDeLaViolencia hashtags.
Jin et al. built their Twitter mentions
networks by extracting more than 20
million Twitter users. Interestingly,
they found the time required for spread
of protest information through such
networks depended on the network’s
substructures. They have also observed
that mass protests emerged swiftly and
sharply on Twitter, i.e. they would be
considered trending topics with short
decline time.
Stochastic models of the kind
described here can have much wider
implications for public services than
just that of model performance. Social
media outlets are being increasingly more
often used as a means of mobilization
and strategic interaction between
participants of mass social movements,
such as protests. While traditionally social
movements have spread through on-the-
ground unions, the use of communication
platforms—such as Twitter and
Facebook—has offered alternative ways
for organization such events.
References
[1] Pastor, L., and Veronesi, P. Uncertainty about
government policy and stock prices. The Journal of
Finance 67, 4 (2012), 1219-1264.
[2] Mörters, P., and Peres, Y. Brownian Motion.
Cambridge University Press, 2010.
[3] Zhou, H., and Lipowsky, R. Network Brownian motion:
A new method to measure vertex-vertex proximity
and to identify communities and subcommunities. In
Computational Science-ICCS 2004. Springer Berlin
Heidelberg, 2004, 1062-1069.
[4] Zhou, H. Network landscape from a Brownian
particle’s perspective. Physical Review E 67, 4
(2003), 041908.
[5] Jin, F., et al. Modeling mass protest adoption in
social network communities using geometric
Brownian motion. In Proceedings of the 20th ACM
SIGKDD International Conference on Knowledge
Discovery and Data Mining, ACM, New York, 2014,
1660-1669.
[6] Guille, A., Hacid, H., Favre, C., and Zighed, D. A.
Information diffusion in online social networks: A
survey. ACM SIGMOD Record 42, 2 (2013), 17-28.
Supplementary Materials
http://github.com/acmxrds/spring-2016
Biography
Marinka Zitnik is a Ph.D. student in computer science
at the University of Ljubljana. She has also completed
research at the University of Toronto, Imperial College
London, Baylor College of Medicine, and Stanford
University. Her interests include machine learning,
artificial intelligence, probabilistic numeric, and
bioinformatics.
Copyright held by Owner(s)/Author(s).
83
 ACRONYMS
ACIS Alan, Charles, Ian and Spatial:
The 3D ACIS Modeler was developed
in the 1980s by Ian Braid, Alan Grayer
and Charles Lang in conjunction with
Spatial Corp. It is a geometric modeling
kernel used in CAD manufacturing,
engineering, and architecture, and has
a C++ architecture that enables robust
3-D modeling capabilities.
CAD Computer-Aided Design: The use
of computer programs to create 2-D or
3-D objects. CAD software is used for
computer animation, special effects,
and to design physical products in a
wide range of industries.
CAM Computer-Aided Manufacturing:
A process that uses computer
software aided machinery and tools
to facilitate the manufacturing
process. CAM is more accurate, more
consistent, more efficient, and faster
than manual machines.
CNC Computer Numerical Controlled:
Tools such as CNC routers and mills are
computer controlled cutting machines
used in computer-aided manufacturing.
A router uses a G-Code, converted
from the interpretation of 2-D vector
drawings or 3-D models via proprietary
software such as CAD.
FDM Fused Deposition Modeling :
A type of 3-D printing, or additive
manufacturing, that is widely used
for creating 3-D objects. Using a
thermoplastic filament that is heated
until its melting point is reached, the
object is then shaped, layer-by-layer.
LOM Laminated Object Manufacturing:
A very affordable and rapid prototyping
3-D printing technique developed by
Helisys Inc. LOM creates a 3-D prototype
using paper or plastic layers that are
laminated or fused-together with
heat and pressure. It is then given the
required shape by cutting it with blade or
laser that is controlled by computer.
SLS Selective Laser Sintering: This is a
3-D printing technique that compresses
tiny ceramic, plastic or glass particles.
The compression, or fusion, of particles
occurs by heating them, using a powerful
laser to create a 3-D object.
84
POINTERS
DIGITAL FABRICATION
Digital fabrication is a type of
manufacturing process in which
machines are controlled by a computer.
According to Dr. Neil Gershenfeld, the
real revolution of fabrication is much
more fundamental than translating
designs into products—it brings
programmability to the physical
world. It draws on the same insights
that led to the earlier digitizations of
communication and computation,
only what is now being programmed
is the physical world rather than
the virtual one. Digital fabrication
will allow individuals to design and
produce tangible objects on demand,
wherever and whenever they need
them. This means a design can be
downloaded and made, reliably and
repeatedly, all over the world without
a maker needing to have specialist
equipment. Offering the prospects of
on-demand mass personalization, with
more localized and flexible production,
digital fabrication is believed by
many to be the start of a potential
manufacturing revolution.
—Tejas S. Khot
THE MANY FACES OF
DIGITAL FABRICATION
“Digital Fabrication in Architecture”
Ryan Hoover teaches an introductory
class on digital fabrication at the
Maryland Institute College of Art,
where students are encouraged to
contribute to a class blog. Andrew
Flanders, a college junior, enrolled
in the course last fall. In this post,
Flanders discusses the role of digital
fabrication in architecture. It should
be noted, while architects are able to
create fascinating designs and models
virtually, the questions and clouds
of doubt regarding the practicality of
those concepts often leave them locked
in the virtual world. Not so when it
comes to digital fabrication. Advances
in digital fabrication technology have
allowed architects to create mind-
boggling designs with relative ease,
which would have likely required the
labor of countless master craftsmen in
times past.
http://dfabclass.com/fall_15/
uncategorized/digital-fabrication-in-
architecture/
“New Materials
for Manufacturing”
By Paul Markillie
3-D printing, or additive manufactur-
ing, is a process of making a three-
dimensional solid object of virtually any
shape from a digital model. 3-D printing
is achieved using an additive process,
where successive layers of material are
laid down in different shapes. Research-
ers in Tennessee have developed an
automated system endearingly known
as BAAM (Big Area Additive Manufac-
turing) for printing car parts. Similar
progress has been made with multiple
academia-industry collaborations. This
article provides a short survey of the
advances in manufacturing techniques,
which could enable large scale produc-
tion in reasonable amount of time.
http://www.economist.com/technology-
quarterly/2015-12-05/new-materials-
for-manufacturing
XRDS • SPRING 2016 • VOL.22 • NO.3
“China’s Revotek Unveils World’s
First 3D Blood Vessel Bio-printer”
Present-day organ transplant
techniques are severely limited and
plagued by long waiting lists, which
can cause delays and, in many cases,
organ rejection by the recipient’s
immune system. The ability to
artificially create a network of blood
vessels will bring us one step closer
to the eventual fabrication of organs,
providing patient specific tissues
almost immediately. According to
the company’s chairman Yang Keng,
Revotek’s 3-D bioprinting system can
now print 3-D stem cells; it includes
a medical imaging cloud platform,
biological inks, a 3-D bio-printer, and a
post-print processing system—a four-
core technological system. With it,
organ rebuilding may become a reality
in the near future.
http://www.3ders.org/
articles/20151026-chinas-revotek-
unveils-worlds-first-3d-blood-vessel-bio-
printer.html
Soft Digital Fabrication
Fab Labs and makerspaces typically
focus on hard-surface object
production. However the first digital
fabrication tool was an electronic
knitting machine dating back to 1976,
which has since been forgotten about
and discontinued. Enter Circular
Knitic, an open hardware project by
artists Varvara Guljajeva and Mar
Canet, which uses digital fabrication
and makers’ tools—like 3-D printing,
laser cutting, makerbeam, and
Arduino—to design an automated and
replicable circular knitting machine.
http://www.varvarag.info/circular-knitic/
XRDS • SPRING 2016 • VOL.22 • NO.3
EMPOWERING THE MAKERS
“What Does One Actually Learn
from Digital Fabrication?”
By Erin Riley
While conceiving an idea and
shepherding it into a tangible form
is significant, it is important to be
able to articulate its value within an
educational setting—revealing the
many layers of learning embedded in
the design-to-making process. Riley is a
FabLearn Fellow at Stanford University,
as well as a contributor to the FabLearn
blog. In this post, she shares her
thoughts on what it means for students
to learn about digital fabriaction.
http://fablearn.stanford.edu/fellows/
blog/what-do-people-learn-using-
digital-fabrication-tools
Fab Lab
Emerging from the MIT’s Center for
Bits & Atoms Fab Lab Program, the
Fab Foundation was formed in 2009
to provide an international network
that focuses on education, service, and
business development. Fab Lab is the
educational arm of the foundation,
serving as a platform for innovation
and invention and providing stimulus
for local entrepreneurship. Fab
Labs are “a place to play, to create,
to learn, to mentor, to invent.” Fab
Labs must provide access to digital
fabrication tools and function to serve
the existing community. All Fab Labs
share common tools and processes;
they must be publicly accessible, free,
and participate in the larger sharing
network that spans 30 countries and
24 time zones.
http://www.fabfoundation.org/
Fab Academy 2016 Tutorials
The Fab Academy teaches principles
and applications of digital fabrication.
It was developed to teach hands-on
skills in Fab Labs around the globe.
It is a fast paced, hands-on learning
experience where students plan and
execute a new project each week. Each
individual documents their progress,
resulting in a personal portfolio
of technical accomplishments. All
course notes, tutorials, and video
lectures in English and Spanish are
archived and can be accessed for free
from their website.
http://archive.fabacademy.org/
archives/2016/doc/index.html
Connect with Other Makers
FabHub is a global directory of digital
fabricators, makers, workshops,
factories, and manufacturers offering
manufacturing services using digital
fabrication technology.
https://www.fabhub.io/
FURTHER READING
“How Sustainable is Digital
Fabrication?”
By Kris De Decker
The process of manufacturing has
been in a state of constant evolution,
and the marginal gains offered by
existing energy saving measures
provide poor incentives to improve
energy efficiency of machine tools at
the cost of hampering productivity.
With billions of people just a click
away from getting factories to work
for them—whether in the cloud or
on their desktops—the population
of producers could be on the verge
of an exponential expansion. This
2014 article from Low-tech Magazine
provides a thorough analysis
questioning the sustainability of this
coming revolution.
http://www.lowtechmagazine.
com/2014/03/how-sustainable-is-
digital-fabrication.html
85
 FEATURED EVENT
DigiFab Con
Microsoft NERD Center
Cambridge, MA
April 22-23, 2016
Want to “print” a bridge or some
Lego bricks? 3-D printing is here
to help. Digital fabrication is not
just limited to 3-D printing though.
DigiFab Con will show attendees
how 3-D printing and other
digital fabrication technologies
impact education, manufacturing,
innovation, design, business,
the arts, and society as a whole.
Various presentations, interactive
demonstrations, panel discussions,
and demos showing digital
fabrication in action are scheduled
for this two-day event. Projects
from all over the world will also be
on display. The conference will be
held at the Microsoft NERD Center.
Friday afternoon’s keynote
will feature Sherry Lassiter, the
director of Fab Foundation; Mike
Adelstein, president and CEO
of Potomac Photonics Inc.; and
Andreas Bastian, who is on the
board of the Enable Community
Foundation and a 3-D printing
scientist at Autodesk.
The cities of Cambridge and
Boston also have many points of
interest for visitors: The Freedom
Trail, Harvard Square, and
tons of art museums. For more
information, please visit http://
digifabcon.org/
—Darshit
— Patel
86
EVENTS
CONFERENCES
Inside 3D Printing
Javits Convention Center
New York, NY
April 10–12, 2016
http://inside3dprinting.com/
new-york/2016/
DigiFab Con
Microsoft NERD Center
Cambridge, MA
April 22–23, 2016
http://digifabcon.org/
RAPID 2016
Orange County Convention Center
Orlando, Florida
May 16–19, 2016
http://www.rapid3devent.com/
FabLearn Europe 2016
Univeristy of Central Lancashire
Preston, United Kingdom
June 19–20, 2016
http://fablearn.eu/
CONTESTS & EVENTS
IAM3D Challenge
The ASME Innovative Additive
Manufacturing 3D Challenge is an
event sponsored by the American
Society of Mechanical Engineers
(ASME). Designed for undergraduate
students, contestants work
individually or in teams to create new
product designs that minimize energy
efficiency. Finalists will present their
work at IDETC/CIE later this year in
Charlotte, North Carolina. The final
day to submit a project is April 15th.
https://www.asme.org/events/
competitions/iam3d-challenge
ProtoHack
ProtoHack is a code-free hackathon
aimed at non-technical entrepreneurs.
During the event, participants use
prototyping tools and mockups to
design and demo an idea, in teams
or individually. ProtoHack events are
taking place in multiple cities this
year, including San Francisco (April
9th ) and Milan (May 7th).
http://protohack.org/
GraDiFab 2016
The first Eurographics Workshop on
Graphics for Digital Fabrication will
be held on May 8, 2016 in Libson,
Portugal. The one-day event features a
keynote session followed by four paper
presentations; the presented research
lies at the intersection between
computer graphics and digital
fabrication technologies. Presenters
will also display physical models of
their work, which will be exhibited
during the workshop. GraDiFab is co-
located with Eurographics 2016.
http://gradifab.org
Machine Learning Summer School
Machine Learning Summer School
(MLSS) is a 10-day event intended
for researchers and professionals.
The school includes lectures from
researchers in both academia and
industry, as well as tutorials focused
on machine learning, data analysis
and inference and related topics.
MLSS will take place at the University
of Cádiz in Cádiz, Spain from May
11–21.
http://learning.mpi-sws.org/mlss2016/
Image by Sean Pavone
XRDS • SPRING 2016 • VOL.22 • NO.3

GRANTS AND SCHOLARSHIPS
Informatics International
Master’s Scholarships
Website: http://www.ed.ac.uk/student-
funding/postgraduate/international/
science-engineering/infomatics
Deadline: April 1, 2016
Eligibility: Academic merit
Benefits: £2,000 tenable for one
academic year
Explanation: The University of
Edinburgh’s School of Informatics
is offering up to 10 scholarships to
none-EU students who are eligible to
pursue a postgraduate master’s degree
in artificial science, cognitive science,
computer science, or informatics.
All candidates must have, or be on
target to obtain, a UK first class or 2:1
honours degree at the undergraduate
level or its international equivalent.
You cannot apply for this scholarship
unless you have already applied
for admission to the University of
Edinburgh.
STEM Majors Scholarships for
Undergraduate and Graduate Students
Website: http://www.afcea.org/
site/?q=foundation/scholarships/
stem-majors
Deadline: April 11, 2016
Eligibility: U.S. citizens majoring in
STEM fields.
Benefits: Tuition, mandatory fees,
and books
Explanation: The scholarship is
sponsored by AFCEA (Armed Forces
Communications and Electronics
Association) for undergrad and grad
students, who have already started
their studies in any STEM field
including: cyber security, intelligence
and homeland security-related
disciplines. Students must be enrolled
at an accredited U.S. college or
university. Applicants may be enrolled
in both on-campus and distance-
Image by Leonid Andronov
learning schools.
XRDS • SPRING 2016 • VOL.22 • NO.3
Google Lime Scholarship Program
Website: http://www.limeconnect.
com/opportunities/page/google-lime-
scholarship-program
Deadline: Will be announced Fall 2016
Eligibility: Undergraduate, graduate,
or Ph.D. students currently enrolled
(2016–2017) at a university in the
U.S. or Canada who are pursuing a
computer science related degree and
have a visible or invisible disability.
Benefits: $10,000 (U.S.), $5,000 (Canada)
Explanation: This Google scholarship
is offered to university studies with a
disability—defined as someone who
has, or considers themselves to have,
a long-term or recurring issue that
impacts one or more major activities,
which others may consider to be a daily
function. Recipients will also be invited
to Google’s headquarters in Mountain
View, CA for the Google Scholars’
Retreat. Candidates will also be
considered for internships with Google.
Xerox Technical Minority
Scholarships
Website: http://www.xerox.com/jobs/
minority-scholarships/enus.html
Deadline: 2016–2017 deadline to be
announced
Eligibility: Minority students enrolled
in a technical science or engineering
degree program at the bachelor level
or above.
Benefits: $1,000–10,000
Explanation: Xerox offers academic
awards to minority students (Black,
Asian, Pacific Islander, Native
American, Alaskan, or Hispanic) who
are pursuing a technical degree, have
a B average or higher, and have tuition-
related expenses that are not currently
covered by scholarships or grants.
FEATURED EVENT
FabLearn Europe 2016
University of Lancashire
Preston, U.K
June 19-20, 2016
Digital fabrication is a
manufacturing process that
uses 3-D modeling software
in conjunction with additive
or subtractive manufacturing
processes. It has seen a massive
rise in popularity with the advent of
affordable 3-D printing technology.
A spin-off of the global
FabLearn conference, FabLearn
Europe is jointly organized by
the Child Computer Interaction
(ChiCI) Group, Stanford University,
the University of Central
Lancashire, and Aarhus University.
It will bring together researchers,
educators, and designers to
explore digital fabrication in
education, making, and hands-on
learning. The conference will also
host various workshops, poster
presentations, and demos.
Attendees willing to travel
should consider the SIGCHI
IDC’16 conference, which be held
in Manchester on the 21st–24th.
There are also various points of
interest, like the National Football
Museum in Manchester and
Avenham Park in Preston. For
more information on FabLearn,
please visit http://fablearn.eu/
—Darshit
— Patel
87
end

BEMUSEMENT

Junk Bonds are Bad Fire

Long-term Investments Island
A man is stranded on an is-
land covered in forest. One
day, when the wind is blow-
ing from the west, lightning
strikes the west end of
the island and sets fire to

http://thisisindexed.com/2016/01/19626/
the forest. The fire is very
violent, burning everything
in its path, and without in-
tervention the fire will burn
the whole island, killing the
man in the process.
How can the man survive
the fire? There are cliffs
around the island, so he
cannot jump off, and there
are no buckets or any other
means to put out the fire.

Laser Scope Find the solution at:

http://xrds.acm.org/
bemusement/2016.cfm

SUBMIT A PUZZLE
Can you do better?
Bemusements would
http://xkcd.com/101/

like your puzzles and

mathematical games
(but not Sudoku).
Contact xrds@acm.org
to submit yours!

Piled Higher and Deeper

http://www.phdcomics.com/synd/c/phd011516xyff.tif

88 XRDS • SPRING 2016 • VOL.22 • NO.3

www.computingreviews.com/best20
 CAREERS at the NATIONAL SECURITY AGENCY

EXTRAORDINARY WORK
Inside our walls, you will find the most extraordinary people doing Computer/Electrical Engineering
the most extraordinary work. Yet it’s not just finite field theory, Computer Science
quantum computing or RF engineering. It’s not just discrete Cybersecurity
mathematics or graph analytics. Information Assurance
Mathematics
Foreign Language
It’s all of these and more – rolled up into an organization that leads Intelligence Analysis
the world in signals intelligence and information assurance. Cryptanalysis
Inside our walls you will find extraordinary people, doing Signals Analysis
extraordinary work, for an extraordinary cause: Business
Finance & Accounting
The safety and security of the United States of America. Paid Internships, Scholarships & Co-op

U.S. citizenship is required for all applicants. NSA is an Equal Opportunity Employer and abides by applicable employment laws and
regulations. All applicants for employment are considered without regard to age, color, disability, genetic information, national origin, race,
APPLY TODAY
religion, sex, sexual orientation, marital status, or status as a parent.

WHERE INTELLIGENCE GOES TO WORK®

Search NSA to Download

www.IntelligenceCareers.gov/NSA