Table Of ContentAmy LaViers
Magnus Egerstedt Editors
Controls and Art
Inquiries at the Intersection of
the Subjective and the Objective
Controls and Art
Amy LaViers Magnus Egerstedt
•
Editors
Controls and Art
Inquiries at the Intersection of the Subjective
and the Objective
123
Editors
AmyLaViers Magnus Egerstedt
Systems andInformation Engineering School ofElectrical andComputer
Department Engineering
Universityof Virginia GeorgiaInstitute ofTechnology
Charlottesville, VA Atlanta, GA
USA USA
ISBN 978-3-319-03903-9 ISBN 978-3-319-03904-6 (eBook)
DOI 10.1007/978-3-319-03904-6
SpringerChamHeidelbergNewYorkDordrechtLondon
LibraryofCongressControlNumber:2013957127
(cid:2)SpringerInternationalPublishingSwitzerland2014
Thisworkissubjecttocopyright.AllrightsarereservedbythePublisher,whetherthewholeorpartof
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation,broadcasting,reproductiononmicrofilmsorinanyotherphysicalway,andtransmissionor
informationstorageandretrieval,electronicadaptation,computersoftware,orbysimilarordissimilar
methodology now known or hereafter developed. Exempted from this legal reservation are brief
excerpts in connection with reviews or scholarly analysis or material supplied specifically for the
purposeofbeingenteredandexecutedonacomputersystem,forexclusiveusebythepurchaserofthe
work. Duplication of this publication or parts thereof is permitted only under the provisions of
theCopyright Law of the Publisher’s location, in its current version, and permission for use must
always be obtained from Springer. Permissions for use may be obtained through RightsLink at the
CopyrightClearanceCenter.ViolationsareliabletoprosecutionundertherespectiveCopyrightLaw.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publicationdoesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexempt
fromtherelevantprotectivelawsandregulationsandthereforefreeforgeneraluse.
While the advice and information in this book are believed to be true and accurate at the date of
publication,neithertheauthorsnortheeditorsnorthepublishercanacceptanylegalresponsibilityfor
anyerrorsoromissionsthatmaybemade.Thepublishermakesnowarranty,expressorimplied,with
respecttothematerialcontainedherein.
Printedonacid-freepaper
PhotousedincoverdesignisbyChristianMoreno
SpringerispartofSpringerScience+BusinessMedia(www.springer.com)
Preface
Severalyearsago,westartedthinkingaboutwhatitmeansforarobottomoveina
particular‘‘styleofmotion.’’Thisquestionaroseoutofadesiretobeabletomake
roboticmovementsmoremeaningfulandrelatabletohumansandacuriosityabout
whatdistinguishesdifferentgenresofmovement.Itquicklybecameclearthatthis
line of inquiry required a whole new set of tools for understanding motion
generation in particular, and control design in general—tools that related to our
perception of human motion and to the performing arts.
In conjunction with the technical development of style-based motions, we
began searching for a community of researchers who were interested inquestions
pertaining to ‘‘Controls and Art.’’ This search resulted in two Invited Sessions at
the American Control Conference (2011 in San Francisco and 2012 in Montreal),
withthesecondonebeingco-organizedwithNaomiLeonard.Wewerepleasantly
surprised by the strong interest in these sessions, and by the many creative
and interesting approaches to using control-theoretic frameworks to investigate,
bolster,andcreateart.Thisbookisalogicalextensionofthetwoinvitedsessions,
where a large subset of our authors contributed papers.
The different chapters in this book fall into one or more of three main camps,
namely research that (1) uses artistic ideas for the purpose of control design and
analysis,(2)usescontroltheoreticideastounderstandandanalyzeart,and(3)uses
control theory as a generator of artistic expressions. It is, however, not our
ambition to paint the complete picture of the research that can be found in the
intersection between Controls and Art, or to imply that this field is close to
maturity.Instead,wehopethatthisbookwillinspirenewresearchinthisexciting
line of inquiry.
Themarriagebetweencontroltheoryandartischallengingbecauseitrequiresa
union between subjective and objective analysis. In fact, while control theory
thrivesunderwell-definedperformancespecificationsandclearmetricsofsuccess,
art thrives on multiple interpretations and evaluations of a single work. As such,
when embarking on an endeavor connecting the two, a shift in perspective is
needed. This book represents such a shift: the list of authors includes artists and
engineers and the list of topics includes dancing robots, swarming dancers,
automated puppetry, reactive museum installments, salsa dancing, intricate
geometric patterns, aerial quadrotor ballets, music generated from math, and
amorphousblobsthatdancetheBhangra.Moreimportantly,weaskourreadersto
v
vi Preface
consider the matters that arise for putting together what may seem like kitschy
combos: How do dancers communicate to execute such highly coordinated
movements? What does it mean for a robot to dance? What do viewers watching
movement notice most? Can we augment human capabilities with technology?
What makes a composition of curves aesthetic? How do people interact with
curatedspaces?Andwhatdotheanswerstothesequestionsteachusabouthuman
creativity?
To answer these questions, the chapters in this book are arranged around
different themes, with the first theme investigating fundamental questions such
as How is movement perceived? (Kingston et al. Chap. 1), What do performers
actually do? (Leonard et al. Chap. 2), and What is in a dance? (Baillieul and
Ozcimder, Chap. 3). The next theme involves the construction of control laws
basedonartisticprinciples,includingaerialrobots(Schoelligetal.Chap. 4),robot
puppets(Jochumetal.Chap. 5),andteamsofmobilerobots(TsiotrasandCastro,
Chap. 6). The last theme turns the relationship between Controls and Art around
by asking how Controls tools can be used in the Arts themselves, including
algorithmic composition using swarm theory (Huepe et al. Chap. 7), automation
for enhancing museum installations (Godbehere and Goldberg, Chap. 8), and the
incorporation of robots in performance (LaViers et al. Chap. 9).
This book would not have been possible without the support from the U.S.
NationalScienceFoundation,theU.S.OfficeofNavalResearch,theU.S.Officeof
Scientific Research, the Swiss National Science Foundation, the Chilean National
CouncilofCultureandArts,andnumerousprivatefoundations,OliverJacksonat
Springer who got as excited about this idea for a book as we were, and Naomi
Leonard who helped connect us to several of the authors here in the Invited
Session she co-organized for the American Control Conference in 2012.
Charlottesville Amy LaViers
Atlanta, October 2013 Magnus Egerstedt
Contents
1 Metric Preference Learning with Applications
to Motion Imitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Peter Kingston, Jason von Hinezmeyer and Magnus Egerstedt
2 In the Dance Studio: An Art and Engineering Exploration
of Human Flocking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Naomi E. Leonard, George F. Young, Kelsey Hochgraf,
Daniel T. Swain, Aaron Trippe, Willa Chen, Katherine Fitch
and Susan Marshall
3 Dancing Robots: The Control Theory of Communication
Through Movement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
John Baillieul and Kayhan Özcimder
4 So You Think You Can Dance? Rhythmic Flight Performances
with Quadrocopters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Angela P. Schoellig, Hallie Siegel, Federico Augugliaro
and Raffaello D’Andrea
5 Robotic Puppets and the Engineering of Autonomous Theater. . . . 107
Elizabeth Jochum, Jarvis Schultz, Elliot Johnson and T. D. Murphey
6 The Artistic Geometry of Consensus Protocols . . . . . . . . . . . . . . . 129
Panagiotis Tsiotras and Luis Ignacio Reyes Castro
7 Generating Music from Flocking Dynamics. . . . . . . . . . . . . . . . . . 155
Cristián Huepe, Marco Colasso and Rodrigo F. Cádiz
vii
viii Contents
8 Algorithms for Visual Tracking of Visitors Under
Variable-Lighting Conditions for a Responsive
Audio Art Installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
Andrew B. Godbehere and Ken Goldberg
9 Style-Based Robotic Motion in Contemporary
Dance Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
Amy LaViers, Lori Teague and Magnus Egerstedt
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
Chapter 1
Metric Preference Learning with Applications
to Motion Imitation
PeterKingston,JasonvonHinezmeyerandMagnusEgerstedt
1.1 IntroductiontoMotion ImitationThroughPuppetry
In entertainment and other artistic endeavors, subjective notions such as “style”
and “aesthetics” play a key role. When control systems are incorporated in these
settings, they are asked to generate behaviors that, rather than achieving a well-
defined physical outcome, serve an esthetic or communicative purpose. In these
situations,theeffectivenessofthecontrolstrategyisultimatelythedegreetowhich
italignswiththesubjectivejudgmentsofhumanobservers.
One specific control application in which this issue has arisen is robotic pup-
petry (see, e.g., [18, 27]), where marionettes with actuated strings—highly com-
plexmechanicalsystems—areaskedtoperformexpressiveandestheticallypleasing
motions in the context of puppetry plays, as in Chap.5, by Jochum et al., and as
shown in Fig.1.1. Given a particular human motion, how should the vastly more
limitedmarionettemovetonotjustmimicthehumanmotionbutalsocommunicate
the same emotional intent? Similar issues are encountered in controls applications
like[8, 16],aswellasininversereinforcementlearningsettings,e.g.,[1, 32, 35].
In puppetry, as in other performing arts, not just any movement is artistically
interestingormeaningful;Whatisthedifferencebetweensomeonejustmovingan
objectandsomeoneanimatingit?Whatisthedifferencebetweensomeonefolding
theirlaundryandsomeonemovingashirtinamannerthatgivesitapersonalityand
bringsittolife?Themovementsofthepuppethavetheintentiontogivethepuppet
B
P.Kingston·M.Egerstedt( )
SchoolofElectricalandComputerEngineering,GeorgiaInstituteofTechnology,Atlanta,
GA30332,USA
e-mail:[email protected]
P.Kingston
e-mail:[email protected]
J.vonHinezmeyer
CenterforPuppetryArts,Atlanta,GA30309,USA
e-mail:[email protected]
A.LaViersandM.Egerstedt(eds.),ControlsandArt, 1
DOI:10.1007/978-3-319-03904-6_1,©SpringerInternationalPublishingSwitzerland2014
2 P.Kingstonetal.
Fig.1.1 Aroboticpuppetexecutingawavemotion
lifeandthroughthosemovementstocommunicateanidea.Itisthemovementofthat
puppetthatistheheartofthecommunication.Butjustwhatisitaboutthemovement
that enables effective communication? Similar questions are pursed in Chap.4 by
Scholleingetal.andChap.9,byLaViersetal.
Toanswerthisquestioninaquantitativemanner,thischapterdrawsinspiration
frompuppetry.And,itisimportanttonotethatpuppetsaredefinedbytheirlimitations
more than their abilities [7, 10, 19]. In fact, the limited movements of the puppet
are a distillation of human movement—so-called motion caricature. Simply put, a
puppetcannotdoallofthemovementofalivingperson,nomatterhowsimpleor
complicatedthepuppetisinitsdesignorconstruction.Itisthereforenecessaryto
distilthemovementthatthepuppetcandotoitsveryessencebyfirstimitatingthe
humanmotion,thensimplifyingittomakeitexecutableonthepuppet,andlastly,
byexaggeratingthemotiontomakeitmoreexpressive.
In order to produce such expressive movements, some guidelines have been
developed through three main classes of puppet movements [33]. The first class
of movements—the primary movements—deals with the overall puppet body and
itcapturesbasicmovementssuchas“up,”“down,”“left,”and“right.”Anyangling
orspinningofthepuppetbodyisalsoconsideredaprimarymovement.“Breath”is
another major, primary movement that the puppet must perform to seem alive. In
fact,mostofthepuppet’semotionsandattitudesareexpressedthroughtheprimary
movementclass,includingthespeedatwhichthepuppetstartsandstopsorwhether
themovementissmoothandcontinuousorjerkyandbroken.
Thesecondarymovementclassisthenextclassofpuppetmovement.Itconsiders
themovingpartsthatareattachedtothepuppet,suchashead,arms,andlegs.The
movementsofthesepartscanmultiplytheemotionsexpressedbythepuppetthrough,
for example, a subtle tilt of the head or various positions of the arms. Finally, the
tertiary movement class concerns puppets that have elements in their designs that
arenotdirectlycontrolledbythepuppeteer.Thesecaninclude,butarenotlimited
to, costume elements such as capes or long sleeves, or hair which can be made of
