Open Archive TOULOUSE Archive Ouverte (OATAO) OATAO is an open access repository that collects the work of Toulouse researchers and makes it freely available over the web where possible.

This is an author-deposited version published in : http://oatao.univ-toulouse.fr/ Eprints ID : 16887

The contribution was presented at CHI 2016 : https://chi2016.acm.org/wp/

To cite this version : Ducasse, Julie and Macé, Marc and Serrano, Marcos and Jouffrais, Christophe Tangible Reels: Construction and Exploration of Tangible Maps by Visually Impaired Users. (2016) In: SIGCHI Conference on Human Factors in Computing Systems (CHI 2016), 7 May 2016 - 12 May 2016 (San Jose, CA, United States).

Any correspondence concerning this service should be sent to the repository

administrator: staff-oatao@listes-diff.inp-toulouse.fr

A

MintrviethTpdinosthcmmvinq

Tangib

1 Un

{firs

Figure 1. a) Th

guided step by

rough (b2) an

ABSTRACT

Maps are essnaccessible toranscribed to n

very expensivessues, we dev

enables visuallheir own, us

Tangible Reelspad that ensuredigital lines tanstructions gui

onto the tablesubsequent exphe points and

confirmed thatmanipulate, andmaps that are validated that nteractions, e

quickly build an

ble ReeM

Julie Du

niversity of TToulouse

st_name.last

he interface ena

y step to correc

d fine (b3) audi

sential in evo visually imnon-editable tae shape changveloped a tanly impaired using a new ty

s. Tangible Rees stability, withangible. In ordide the user to e and create ploration, the d lines that tht Tangible Rd that visually

built with ththe designed

enables visualnd explore map

ls: ConMaps by

ucasse1, Mar

Toulouse – Ie, France t_name}@ir

ables a visually i

ctly place each T

io guidance inst

receiving a

veryday life, mpaired users.

actile graphicsging displays. ngible tabletopsers to build taype of physi

eels are compoh a retractable

der to construcprecisely placelinks betweendevice providehe user touche

Reels are stabimpaired users

hem. A followsystem, inclu

lly impaired ps of various c

nstructiy Visua

rc Macé2, M

RIT

rit.fr

impaired user t

Tangible Reel. T

tructions, and p

auditory feedba

but inherentThey must b

, or rendered oTo tackle thep interface thangible maps ocal icon call

osed of a suckreel that rende

ct a map, aude Tangible Reen them. Durines the names es. A pre-studle and easy s can understan

w-up experimeuding non-visu

participants complexities.

ion andally Imp

Marcos Serra

to render digita

To construct a l

presses the suck

ack when pointi

tly be on

ese hat on ed

ker ers dio els ng of dy to nd ent ual to

Autho

Non-vi

INTRO

In edupeopleand mTechnoorientainherenbe traobservare tragraphictime-cographic[10]. Madjustavisuall

These of grainteracaudio-tand/or based displayexpens

d Explopaired U

ano1, Chris

2

To{first_nam

al maps tangible

line, the user at

ker pad (b4). c)

ing at lines and

r Keywords

isual tangible i

ODUCTION

ucational cente, graphical repmaps are widology, Engineeation and mobintly inaccessib

anscribed into ved usages in aanscribed into cs. The produconsuming proccs specialist [Magnetic mapable, but they ly impaired use

issues can be aaphics or mactive. Howevertactile graphichaptic graphicon a single p

ys are very sive [29].

oration oUsers

tophe Jouff

CNRS - IRIoulouse, Franme.last_nam

e by using Tang

ttaches two Tan

The user can ex

pads.

interfaces; inte

ters for blindpresentations

dely used in ering and Mathility lessons [16ble to visually i tactile grapha specialized Iraised-line, h

ction of the ficess that need30], and resulps are easier cannot be con

ers without ass

alleviated by auaps (see [30]r interactive gcs cannot be cs (see [31]) upoint of conta

promising b

of Tang

frais2

T nce

me}@irit.fr

gible Reels. b) T

ngible Reels (b1

xplore the tang

eractive graphic

d and visuallysuch as chartsSTEM fields

hematics), but a6]. As visual gimpaired usershics. In this Institute: visua

hand-crafted orirst two types s to be done blts in non-edit

to produce nstructed or acsistance.

utomating the ) and/or mak

graphics are aledited [2], ansually enable ect only. Shape

but are still

gible

The user is

1), follows

gible map,

cs;

y impaired s, schemas (Science, also during

graphics are , they must study we

al graphics r magnetic relies on a

by a tactile table maps and more

ccessed by

production king them so limited:

nd acoustic exploration e-changing extremely

WvrtaavddinpplincwtaprabccRecnimacclictado

Toadesrvthdcin

O

Twthatouem

We aimed to visually impairrepresentationsabletop interfa

as they providevia the manipdefined in [9]),digital informnterface for th

people has prevprovided a limiimited number

non-visual tangcost device thwithout any aangible any di

points and lineraised-line mapand explorationby-doing”), wicontact explorconstructed a nReels. A Tangensures stabilitcreate physical non-visual tabmpaired user t

and to interaconstruction, acan place the inked to anoth

can be construactile explorat

digital points anon their physica

This paper descon observationsa specialized design processevaluation constable, easy representationsvisually impairhat enables a v

digital maps consisting in ncreasing com

OBSERVATION

The term “grapwhose spatial lhe reader: dia

are particularlyo-day life, bu

understand andelements. The vmap, itinerarie

overcome thred users with of visual gra

aces are, in the a way to intulation of ph and therefore

mation into a he exploration oviously been iited accessibilir of points, it wgible interfacesat is accessib

assistance, andigital graphica

es. It should cops) because it n of graphics (ithout the resration. In thinovel type of ible Reel is coty, and of a rlinks between

bletop interfato build a tangact with it (audio guidance

phicons correher with the structed betweention, the user nd lines by peral representatio

cribes three ms of the use aneducational I

s of Tangiblenducted to ch

to manipula built with thred user; 3) wvisually impair

tangible. Weconstructing a

mplexity.

N: USAGE OF

phics” brings tlayout is used gram, figure,

y interesting asut also becausd mentally intevarious types oes, etc.) clearl

hese limitationh a way to coaphics on theiris sense, partiteract with digicons (i.e. phalso provide aphysical for

of graphs by vimplemented [ty to mathemat

was a rousing fs. We intendedle to visually

d that allows al content that omplement exi

allows dynam(with the bene

striction of a is study, wephysical icon omposed of a etractable reel

n phicons. We ace that enabible map using(see Figure e is provided ectly. As eachring of a retrac two Tangiblecan listen to t

rforming a tap ons.

ain contributiod production onstitute; 2) we Reels, and

heck that Tante, and thathem can be u

we introduce ared user to indee report on and exploring

F TACTILE MA

together a varito provide co

drawing, map,s they are essense they requiregrate the spatof maps (metroly illustrate h

ns by providinonstruct physicr own. Tangibcularly relevangital informatiohysical icons, a way to translarm. A tangibvisually impair18]. Although tical plots with

first step toward to design a lo

impaired usethem to rendis composed

isting tools (e.mic constructioefit of “learnin

single point e designed an

called Tangibsucker pad th

l that is used also developedbles a visualg Tangible Ree1). During thso that the us

h phicon can bctable reel, line Reels. Durinthe name of thand hold gestu

ons: 1) we repoof tactile mapswe describe th

report on thgible Reels at the physicunderstood by a novel interfaependently mak

its evaluatiog four maps

APS

iety of materiaontent or data , etc. [27] Mantial in our dare the reader ial layout of tho map, overviehow diverse th

ng cal ble nt, on as

ate ble ed it

h a rds ow ers der of .g. on

ng-of nd ble hat to

d a lly els he ser be

nes ng he

ure

ort in he he

are cal

a ace ke

on, of

als to

aps ay-to he ew he

complechose t

To betspecialgeogramobilitgeograCESDVprovidimpaireWe idand han

Figur

orie

Raisedthis InpaperThermhand-cheated and peon a sand cre

Each tedependstudentdifferedisplayusuallygiven mare prsince versionof textuproduc

Magnestudentitinerarrepresemagnestands route, tthe teaMagneconfigufor exa

exity of tactilto focus on the

tter understandlized educatiaphy lessons wty (O&M) less

aphy teacher V-IJA Institutes education ed people, fro

dentified three and-crafted.

re 2. a) A raised

entation and mo

r

d-line maps arenstitute, they ar

techniquesmoforming conscrafted mold m

in a vacuum, ermanently defspecial heat-seeates relief.

eacher owns diding on the agts. During a

ent maps as thyed on a tactiley do not includmap. This numrogressively ina tactile map

ns of the same ures (1-3) each

ce around 1200

ets are more ot learn an itinery for the stuentation on a et is placed on

for and lets tthe students macher can cheets are also urations (road ample).

e graphics cae construction a

d how and whyional centers

with four studeson with one st

and two O&te in Toulouseand training f

om early childtypes of map

d-line map. b) A

obility lessons.

represent itiner

e mainly used bre made with but other

sists in placingmade of differen

the plastic sheformed. Swellnsitive paper

ifferent maps wge, ability and

lesson, teachhe amount of e map is limite

de more than fimber can be incntroduced to np cannot be map are used

h time. In the I0 tactile docum

often used by Oerary (see Figuudents, the tea

magnetic bothe board, the

the student toumay be asked to

eck whether iused to repcrossing or co

an be. In this and exploration

y tactile maps as, we obserents, one orientudent, and int&M instructoe, France. Thifor around 12

dhood to later ps: raised-line

A wooden mode

c) Magnets are

aries.

by geography tthermoformingtechniques e

g a sheet of plant textured papeet is shaped bl-paper maps athat swells wh

whose content visual impairm

hers often neinformation th

ed, but also beive different texcreased when thnew textures. readily edite

to introduce a Institute, four t

ments each year

O&M teachersure 2c). To prepacher builds a oard: every tim

teacher indicauch it. After wo rebuild the roit has been u

present complonfiguration of

study, we n of maps.

are used in rved two ntation and terviewed a ors in the is Institute 20 visually adulthood. , magnetic

el used for

e used to

teachers. In g or swell-exist [4].

astic upon a pers. When y the mold are printed hen heated

may differ ment of the eed to use hat can be ecause they xtures on a he students

However, ed, several

limited set transcribers r.

s to help a pare a new simplified

me a new ates what it walking the oute so that understood. lex spatial f a building

We also observed a number of hand-crafted graphics made out of rope, wood, felt or cardboard. They represent geometrical concepts (shapes, open/close features, parallel/perpendicular) and O&M elements (different types of crossings and a map of the school neighborhood for example).

These observations highlight one major issue: when a map is required, it has to be “materialized” with the assistance of a sighted person, which is a time consuming process. The students cannot access digital maps immediately and independently. Furthermore, the maps that are produced this way are not interactive and not editable, which limits the way users can interact with them and hence their autonomy.

To be fully accessible for visually impaired users, maps should be available without assistance and instantly. In addition, they should be interactive and editable, so that they could support dynamic operations such as zooming, panning, annotations, as well as other advanced functions (e.g. computing distances).

RELATED WORK ON INTERACTIVE MAPS

To alleviate the aforementioned issues, different approaches relying on new technologies have been used. Zeng and Weber [31] classified the different types of interactive maps in four categories, depending on the device and interaction used. Virtual acoustic maps use verbal and non-verbal audio output to render geographical data. For example, Zhao et al. [33] presented thematic maps explored with a keyboard or a tablet and producing string pitch and spatial sound. Virtual tactile (or haptic) maps most often rely on a force-feedback device. For instance, SeaTouch [26] enables visually impaired users to explore a maritime environment relying on haptic feedback, sonification and speech synthesis. The BATS [20] or HapticRiaMaps [11] are other examples of virtual tactile map using force-feedback devices. TouchOver map [22] provides visually impaired users with a basic overview of a map layout displayed on a mobile device through vibrational and vocal feedbacks. Kane et al. [13] described complementary non-visual interaction techniques that allow finding a target on large touch screens. Audio-tactile maps consist in a raised-line paper map placed over a touch-sensitive screen that provides audio descriptions of tactile elements when touched (see [19] and [30]). In contrast to virtual acoustic and tactile maps, these maps provide multiple points of contact (potentially all the fingers), and proved to be usable for learning spatial configurations [2]. Finally, Braille maps displayed on refreshable displays are a promising approach to create interactive maps. Zeng et al. [32] used a BrailleDis 9000 tablet device consisting in a matrix of 120x60 pins that can be moved up or down. Their prototype allowed visually impaired users to explore, annotate and zoom in or out. Similarly, Schmitz and Ertl [23] used a HyperBraille to present different types of maps representing buildings or large outdoor areas. The main drawback of the virtual maps

is that they provide a single point of contact (e.g. a phantom device), which forces the user to explore the map sequentially, and mentally integrate a large amount of information through space and time. However, they do not require a raised-line map overlay and can theoretically allow panning, zooming, and dynamic updating. Refreshable displays can provide both multiple fingers exploration as well as dynamic update, but these devices are extremely expensive, and hence relatively unusual.

Tangible maps for the visually impaired may present a number of advantages: they could be built autonomously using appropriate feedback, which may support learning-by-doing activities, provide multiple fingers exploration and allow dynamic updating while being affordable.

Towards tangible maps

A number of tangible user interfaces have been developed to enable sighted users to interact with a map. GeoSpace [9] is an interactive map onto which objects are placed. Their location modifies the digital map position and extent. Urp [28] allows urban planners to simulate wind flow and sunlight, and to observe their consequences on physical building models placed onto the tabletop. With the MouseHouse Table [8], users can model several arrangements of urban elements such as streets and buildings by placing paper rectangles on the table and visualizing the behavior of pedestrians.

Two devices have been specifically designed for visually impaired users. The Tangible Pathfinder [25] allows them to construct a map using small objects that represent pavements, sidewalks, etc. Audio instructions and feedback assist the user in placing the objects and exploring the map. Schneider et al. [24] designed a prototype for route construction by telling the user the length of building blocks and where to place them on a magnetic board. These devices are devoted to route or neighborhood exploration, and can hardly be adapted to other types of graphical content. In addition, to our knowledge, they have not been formally evaluated, and the construction of a tangible map by a visually impaired user on its own has not yet been demonstrated. In this study, we designed and evaluated a tabletop tangible interface that enables a visually impaired user to construct and explore different types of maps, with different levels of complexity.

The current work is in line with two other research projects [17][18]. In [17] the device provides visually impaired users with multimodal feedback to accurately place objects (called TIMMs) in order to create and modify graphs and diagrams. The authors suggest that a tactile line could be added between two TIMMs with a piece of yarn for example, but did not indicate how the user would select the correct length and could interact with the line. In [18], the device allows the exploration of line graphs and bar charts. Phicons are placed in a restricted physical grid (9x7 cells) in order to represent the top of a bar or the turning point of a linear function. Relying on an evaluation with four users,

thkfv

TptorbppdtaotofscoppIco

M

IthTpac

D

FgfaIunIs(sm

D

Tedaddliliu[

he authors oknocked over dfew recommenvisually impair

This review opoints. First, alto address som

research projecby visually impprototypes descpunctual symbodigital contentangible as we

other graphicalo design phico

fully addressedstable but maycamera placedoccluded by uphysical grid potential locatiIn addition, sconstruction ofonly points.

MAKING DIGIT

In this section,hat should be

Tangible Reelphysically linkand lines), anconstruct and e

Digital and phy

First of all, it graphics is mafine details ofadaptation procIn our system, using a simple name and two ID, a name andsystem does no(such as mobilsparse spatial maps.

Design of the T

To build the phessential to prodifferent lengtappropriate toodifferent lengthdifferent lengthimits the numikelihood of e

used in the fiel14][1][21]).

observed that during the expldations concered users.

of the literaturthough tangiblme limitationscts about autonpaired users arcribed above hols tangible (i.). Obviously, ll because thel representationons that are stad yet. The ma

y be unreliabled above the tausers’ hands. Mthat holds th

ion, but still osuch a solutif most graphi

TAL MAPS TA

, we describe e made with ts (they are t

ked to each othnd the interacxplore maps.

ysical maps

is important tandatory for tf an outline mcess and curvemaps are definsyntax: each pcoordinates; e

d the IDs of stot aim at the lity maps), butconfigurations

Tangible Reel

hysical represenovide the userths. Retractabol as their st

hs. This is morhs (e.g. pre-cut mber of requerrors. Retractld of HCI rese

the objects loration. Hencrning the desig

re highlights le user interfacs of interactivnomous constre seldom. Seco

have only focuse. physical andit is essentia

y are mandatons. Third, the able and reliabagnetic board . The objects aabletop, whoseMcGookin et e objects, wh

observed uninteion is not recs that includ

ANGIBLE

the characteristhe system, thtracked by thher to represection techniqu

to know that sactile exploratmust be remoes are most oftned by a set ofoint is associat

each line is asstarting and endconstruction ot rather at thes such as ove

s

ntation of a digr with a way tle reels appetrings can bee usable than uwool yarn sucired steps antable reels havearch as input d

were regulare they provide

gn of phicons f

three importaes are promisin

ve tactile mapruction of maond, the last twsed on renderind associated tol to make linory in maps anquestion of hoble has not beused in [24]

are tracked bye view may bal. [18] used

hich limits theentional knockelevant for th

de lines and n

stics of graphihe design of thhe tabletop, annt digital poin

ues designed

simplification tion [4]. Hencved during th

ften straightenef points and linted with an IDsociated with ding points. Thof detailed mae construction erview or met

gital map, it wto draw lines eared to be e pulled out using material ch as in [17]) annd therefore thve already bedevices (see e.

rly e a for

ant ng ps, aps wo ng

o a nes nd ow en is

y a be

a eir ks. he

not

ics he nd nts to

of ce, his ed. nes , a an he

aps of

tro

was of an to of nd he en .g.

Accordobjectsexploreover. Twas apidentifiReels. stable systemobjects

Figure

Tangib

can be

infrar

We trieglue doelectrofirst twobjectsfilled wFigure

Weight

Weighweighttests wcm higadequasilicontrack tThe bathick cthen gpassed(1cm. h

Sucker

Suckerthe tabpressedand 2 cunder wglued oby pull

Add-on

Reels also apto andreels w

ding to [18], is to the table e the map wit

This was even mpplied to the

fied three requiThey had to beduring explora

m; as small as s that can be pl

e 3: Weights (a)

ble Reels. c) A t

e tracked by a c

red frame (highl

ed a number oots, adhesive t

ostatic screen pwo requirements fulfilling thewith lead calle 3).

ts

hts were inspiret” of the phic

with a visually igh and 4 cm wate. To furtherne O-ring was the objects waase and the tocardboard stronglued to the tod through a hohigh) to keep i

r pads

r pads can easibletop screen, d. We used prcm high once cwhich a tag caonto the top. Tling on a small

ns

can be harmfupply a strong td make them with a lock/unlo

it was also esso that a visuthout moving more importanobjects by th

irements for the: easy to movation; identifia

possible to mlaced onto the

) and Sucker pa

tag is positioned

camera placed u

lighted in blue)

during explora

of adhesive matape, Wikki Stiprotectors but nts. Finally we e requirementsed Weights, an

ed by [18] whocons to ensureimpaired user,wide cylinder r improve adheadded under t

as placed insidop of the cylinngly glued to thop of the cylinook fixed at thit close to the ta

ily slide along and strongly sofessional flat compressed) than be attached (The sucker padl strip that exte

ful when the stension to the move. Theref

ock button. Be

ssential to firmally impairedthem or knoc

nt in our case ahe retractable he design of thve during constable and trackamaximize the tabletop.

ads (b) are two

d underneath so

under a plate gl

) detects the use

ation.

aterials such asicks, anti-slip gnone of them fuend up with tw

s: small plasticnd flat Sucker

o suggested “ve stability. Af we found thatwith 180 g o

erence of the Wthe base. The tde this ring (Fnder were madhe cylinder. Thnder, and its e bottom of thabletop.

a smooth surfastick to the scsucker pads (

hat present a lar(Figure 3c). Thds can easily bends from its ba

string retracts.objects they ar

fore we used sides, to make

mly set the user could

cking them as a tension

reels. We he Tangible truction but able by the number of

designs of

o that they

lass (d). An

ers’ fingers

s Blu-Tack, gel pads or fulfilled the wo types of c cylinders r pads (see

varying the fter several t filling a 6 f lead was Weights, a tag used to Figure 3c). de out of a he reel was string was

he cylinder

ace such as creen when (4 cm wide rge surface he reel was

be detached ase.

They can re attached retractable it easier to

link two objects, we fixed a strong neodymium magnet at the extremity of the reels’ strings, and added a metallic bracelet to the objects. The bracelet was wrapped around the bottom part of the cylinder for the Weights and around the reels for the Sucker pads.

CONSTRUCTION AND EXPLORATION TECHNIQUES

Tangible Reels are placed next to the user, on the bottom side of the table. Audio instructions and feedback are provided so that the user can gradually construct a simple physical representation of the map by placing the Tangible Reels (see Figure 3d). During exploration, the user can retrieve the name of the points and lines using finger interactions. All the values mentioned afterward (distances and timers) were based on observations made during preliminary tests.

Constructing the map

Construction instructions

Each line is constructed using two Tangible Reels attached to each other. Three instructions indicate to the user what is the next action to perform (see Figure 1):

“New object”: at the very beginning of the construction, and each time a new line has to be built, the user has to place a Tangible Reel on the table. As soon as the Tangible Reel is detected, guidance instructions are provided.

“Attach an object to the right/to the left/below/above”: To construct a line the user has to pull out the string of a new Tangible Reel and attach it to the metallic bracelet of the last one that was placed.

“Attach an object to <name of the object>, to the right/to

the left/below/above”: The start point of the line to be built is not always the last Tangible Reel that the user has placed. In this case the system gives the name of the object to which the new Tangible Reel must be attached to.

Feedback

“Attached”: This instruction is played when the system detects that the new Tangible Reel is close enough to the one that it must be attached to, and is immediately followed by guidance instructions.

“<Name of the point> found”. The system informs the user when the Tangible Reel is at the right location by giving the name of the point represented by the Tangible Reel. If the Tangible Reel is the end point of a line, the instruction “<name of the line> built” is played.

“Object lost”: The user is informed when the Tangible Reel that is being moved has not been detected by the system for more than 2500 ms.

The last instruction is repeated every 7000 ms until the appropriate action is done by the user. When the “attach an

object” instruction is repeated, the name of the object to which the user must attach a new Tangible Reel is also given.

Guidance instructions

Depending on the distance between the Tangible Reel that the user is currently moving and the position of the target

point, two types of guidance instructions are provided: rough guidance (every 3500 ms) and fine guidance instructions (every 1500 ms).

Rough guidance instructions (Figure 1b, step 2). When distance is superior to 15 cm, the system indicates the direction of the target (up / down / left / right / up and right / down and right / up and left / down and left) as well as the distance in centimeters. This enables the user to either quickly slide or lift the object towards the target.

Fine guidance instructions (Figure 1b, step 3). When the distance to the target is inferior to 15 cm, the system provides more frequent feedback to indicate the direction to follow (up / right / down / left). As long as the target has not been reached, the system repeats the procedure.

Exploring the map

When exploring the map, the user can listen to the name of a point or a line by performing a tap and hold gesture above it. To avoid unintentional selections, the user must select one point or line at a time for at least 700 ms (see Figure 1).

IMPLEMENTATION

Hardware

Our tabletop was a 100 x 100 cm plate glass. The setup also included a projector to illuminate the surface and a webcam to detect tagged objects. Both were placed beneath the plate glass. A multitouch IR frame was placed two centimeters above the plate glass (Figure 3d) in order to detect the fingers. To achieve a high quality of tag detection, we restricted the area of work to 80 x 57 cm. The projector, webcam and IR frame were connected to a laptop.

Software

The Tangible Reels were tracked using the TopCodes library [7], which allowed using small circular tags that fit under the objects. The IR frame sent messages containing the finger input state (pressed, updated or ended) and position using the TUIO protocol [12]. We used the MultiTouch4Java library (MT4J, [15]) to receive TUIO messages, and to display the image of the map when needed (e.g. for debug) as well as the position of the physical objects and lines. Audio instructions were provided with a SAPI4 compliant Text-To-Speech engine distributed as part of the CloudGarden TalkingJava SDK 1.7.0.

PRE-STUDY: TANGIBLE REELS USABILITY

The aim of this pre-study was to investigate whether the two types of Tangible Reels were stable and easy to manipulate, but also to verify that built tangible maps were understandable by visually impaired users. It was done for testing the object design only and was performed without any interactive instruction or feedback. We used two types of maps that are frequently used by visually impaired users: metro maps and overview maps (Figure 2a). The Braille Authority of North Canada defined overview maps as maps that “may not have specific detail that would allow some readers to plan a walking route, but instead are designed to familiarize and orient the reader with the area encompassed”.

P

Wm1eSptwcathfthtoruucRrrmcr

E

Wv

d

mmoadoo

P

Ttweooqaoa

Figure 4: A set

(a, b) and for t

metro map

Participants and

We recruited fmales) aged be14.9). The stuexploration tasSucker pads anparticipants wewo objects to

comfortable. Tand one overvihe experiment

four minutes tohe explorationo do it as accu

rather than the used a Dycem used to Dycemconsisted in reReels, as quickraised-line mreconstruction. memorization constructing thraised-line map

Experimental d

We used a wivariables for ta

Object desigdescribed below

Map (M). Fmaps (see Figumetro map andother representand required twdesigned two objects) and twobjects).

Procedure

The study was wo designs. A

exploration taskof these tasks overview mapquestionnaire. according to thorder of the blalso counterbal

of adapted map

the construction

s while (b) and

d tasks

four legally blietween 31 andudy consisted sk and one cond twelve Weigere told how toogether. They ask 1 consisteiew map that ter. Participano explore those

n, they had to durately as possdistances. To

sheet of paper,m paper, used econstructing kly as possibl

map and hadNo time l

or the conshe tangible map again.

esign

ithin-subjects sks 1 and 2 (O

gn (O). We evaw: the Sucker p

For the explorure 4): two red required ninted an overviewelve objects. other fictive

wo other over

made up of twA block consist

k and finally ta metro map

. After each They also

heir preference ocks was counlanced the two

ps used for the

n task (c, d). (a)

(d) are two ove

ind persons (twd 65 years (M

in one trainonstruction tasghts. During tho construct a lcould practiced in exploringhad been prev

nts had respecte maps, and im

draw the map. Tsible, focusing

draw the map, and one subjemagnets on a

two maps wile. Participantsd to memorimit was im

struction. Oncap, they could

design with tbject design an

aluated the twpads and the W

ration task weepresented a pne objects to bw map (inspirFor the constmetro maps

rview maps (r

wo blocks correted in training the constructiop was presenttask, participaranked the at the end of

nterbalanced amo sets of maps,

exploration tas

) and (c) are two

erview maps.

wo females, twM = 48.2, SD ning phase, onsk using twelv

he training phasline by attachine until they feg one metro maviously made btively three anmmediately aftThey were askon the topolog

ps, three subjecect, who was na board. Task th the Tangibs were shownrize it befo

mposed for thce they start

not explore th

two independend Map):

o object desigWeights.

e designed fopart of a fictivbe built and thred by Figure truction task, w

(requiring ninrequiring twelv

esponding to thfollowed by th

on task. For boted and then ants answered

object desigthe session. Thmong users. Weach containin

sk

o

wo =

ne ve se, ng

felt ap by nd ter ed gy cts not

2 ble

a ore he ed he

ent

gns

our ve he 2)

we ne ve

he he

oth an

a ns he

We ng

two mconstru

Measu

For thobject order tthe hannot invthat hamagneDycemcorrect0/10 mmeans judges distancof draw

Result

In thisParticip

Objects

The avof the was 0.2= 0.01)SuckerknockeWeigh

Map dr

Figure given w= 2.7) for themarks,(SD =construmodel.the wh

Figur

(a) and

Questio

The palevel odisagreobject with th

frustra

maps for the uction.

ures

he exploration displacementto evaluate sund-drawn mapvolved in the pave been explets (Participantm sheet. We tness of the dr

means that the tthat the two mto focus on th

ces. We had pwings that shou

ts

s pre-study, wpants are later

ts Stability

verage distanceobjects before28 cm (SD = 0) for the Sucker pad got deted over three

hts.

rawings and m

5 shows examwere 8.1 (SD =for the overvie

e overview dra, the average

= 1.4). One mucted by the . The subjects

hole raised-line

re 5: two drawin

d the overview m

respectively

ionnaire and ra

articipants speof agreement oee and 7 = strdesign. The f

hese objects is

ating. The la

exploration a

task, we meaduring explora

ubjects’ spatials to four indepproject, alongslored. The mat 4) were thor

asked the jrawn maps as two maps were

maps were highhe topology o

previously showuld receive 0, 5

we mainly focureferred as P1

e in centimetee and after the 0.03) for the Wer pads. Duringtached; none

Weights, and

map construction

mples of draw= 1.3) for the ew maps. P4 o

awings (1.5 andmark for the

metro map anparticipants windicated that

e map.

ngs made after

map (b) present

y given a score

anking

ecified on a 7on a series of rongly agree),first four items: 1) pleasant;

ast item wa

and two map

asured the unation (distancel learning, wependent judgesside pictures oaps that were roughly reprodjudges to evcompared to

e not similar athly similar. Wef the map rathwn them three5 and 10.

used on qualit, P2, etc.

ers between theexploration o

Weights and 0.0g all the explorwas moved.

d P3 knocked

n

ings. The avermetro maps an

obtained the lowd 2.5). If we eoverview map

nd three overvwere not simithey could not

exploring the m

ted in Figure 4.

of 8.6 and a 6.6

7-point Likert statements (1 for each task

ms were: Build

; 2) difficult;

as: 5) it is

ps for the

nintentional e in cm). In e presented s who were f the maps made with

duced on a valuate the

the model: t all; 10/10 e asked the her than on e examples

tative data.

e positions f the maps 08 cm (SD rations, one P1 almost

d over two

rage marks nd 5.7 (SD west marks exclude her ps was 7.0 view maps ilar to the t remember

metro map

. They were

6.

scale their = strongly

k and each ding a map

3) fast; 4)

easy to

u

ea

pAetaWp

Q

CthththpcdpwkeoP

ArPth“PPli

C

BvcmmacWdapdup

unintentionally

exploration taskallowed me to

percentage of aAll the particexploration taskask (two prefe

Weights). Ovepads as a globa

Table 1. Perce

each item and

construction ta

Qualitative feed

Concerning thehey had to rephe Weights thahey could be

position when concerns whendeclared that paying attentiowas reported knocking themexploration. Twof the Weights P3).

As for the Sucreduced height P3, P4), and twhem during th

“the advantage P3 described thP4 said that “aittle bit annoyi

Conclusion to

Both Tangible visually impairconditions, onemap with themajority foundappeared to bcoherent with tWeights were drawback of thas easily as theparticipants haduring the counlikely to hpositioning of t

y move or knoc

k, a sixth itemunderstand th

agreement for eipants preferrk but results werred the Suckrall, three par

al best choice.

ntage of subjec

d Object Design

asks. The darke

dback

e Weights, twoplace several oan the Sucker peasily moved,constructing t

n exploring thhe “missed o

on not to knockby P1 who m over” anwo participantswas an issue,

cker pads, thrallowed them

wo stated that e exploration (is that they do

he sucker padsattaching and ding”.

the pre-study

Reels proved ed user. Over te participant oe Tangible Red it pleasant.

be more stablethe fact that tweasy to unin

e Sucker pads e Weights. Hoad to remove tnstruction tas

happen in a the object is pr

ck over these

m was evaluatedhe maps”. Tabeach statement.red the Sucke

were mixed for ker pads and thticipants prefe

cts who answere

n, after the explo

er the cell, the h

o participants sobjects it was epads (P2, P4). , it was handythe map. Parti

he map with tone object bek them over”. said that he

nd that they s also reportedrather than the

ree participantsto better explothey were few(P1, P4). P4 alo not take a lots as “cool”, “ldetaching the S

y

to be easy to the four Objectnly consideredeels was diff

However, the than the Wwo participantntentionally knis that they canwever, it shouthe Sucker padk in this pre

real scenarirecisely guided

objects. For thd: “these objec

ble 1 shows th. er pads for ththe constructio

he other two therred the Suck

ed a 5, 6 or 7 for

oration and the

higher the value

stated that wheasier to do wiP3 stated that

y to adjust theicipants reportthe Weights: Pecause [he] w

The same issuwas “afraid

hindered thd that the heigeir diameter (P

s said that theore the map (P

wer risk to movlso declared tht of place” whiight” and “funSucker pads is

manipulate byt Design * Tasd that buildingficult while thhe Sucker pa

Weights. This ts found that thnock over. Onnnot be remov

uld be noted thds several time-study. This io because thd before fixatio

he cts

he

he on he

ker

r

e

e.

en ith as eir ed P2

was ue of he

ght P1,

eir P1, ve

hat ile n”. s a

y a ks

g a he

ads is

he ne ed

hat mes

is he

on.

SeveraWeighknocketo be construpads bwere gthese oWeighfollowi

The mconstrumemorparticipundersdrawindifficucompledifficusuggesbe tooinvestiof Tan

STUDY

The aiwhole investithe madescrib

Partici

We recaged bborn bbetweelight pcould n

Experi

We deFigure 8, 10, 4), as weach mlines athree dM8, Mwith a

Figu

using T

al participants hts hindered ted some Weigh

hindered bucting a map.better meet theglobally preferrobservations,

hts were not using experimen

marks attribuucted with Trized by visuapants found th

stand the mapngs showed ulties, especiallex. Indeed, thre

ult to understasts that the mapo complex. In igated the effe

ngible Reels.

Y: MAP CONS

im of this studinteractive

igate whether aps led to usabed in the Impl

ipants

cruited 8 legalbetween 24 andblind; one becen 4 and 6 yearperception but not perceive an

imental design

esigned four 6) by graduall

12), lines (5, 6well as crossin

map contained as well as one pdifferent lines.M10 and M12.

numerical labe

ure 6: the four m

Tangible Reels.

also reportedthe explorationhts over. It is e

by the object To sum up, e stability reqred by three peven though w

sable, we chosent.

uted to drawangible Reels

ally impaired uhat the Tangibl. However, ththat some ply on overviewee participantsand and memps built with T

the follow-upect of map com

STRUCTION A

dy was to evaldevice with increasing theability issues. lementation se

lly blind persod 65 (M = 43.8came blind befrs old and one were unable t

nything at all.

n

maps of diffly augmenting6, 7, 8) includings between li

two horizontapoint that wasThese maps ar. All points anel, ranging from

maps that the p

The complexity

right.

d that the hein, and two pessential for thts when expit appeared th

quirement, andarticipants. Acwe do not cone the Sucker p

wings show ts can be expusers. Three ole Reels allowhe existence oparticipants ew maps that w

s said that they morize. This oTangible Reelsp study, we smplexity on th

ND EXPLORA

luate the usabiSucker pads

e levels of comWe used the

ction above.

ons (2 females8, SD = 14.4).fore the age oat 16. Three h

to discern shap

ferent complexg the number ofing oblique linines (0, 2, 4, 6al lines and tws the start or enre later referrednd lines were m 1 to 12.

articipants had

y is increasing f

ght of the participants he users not ploring or hat Sucker d that they ccording to nsider that

pads for the

that maps plored and out of four ed them to f incorrect xperienced were more were quite

observation should not

specifically he usability

ATION

ility of the s, and to mplexity of e apparatus

s, 6 males) Four were

of 1; three had residual pes; others

xities (see f points (6,

nes (1, 2, 3, 6). Besides, wo vertical nd point of d to as M6, associated

d to build

from left to

WmAo

I

TbDmincm

Aotyo

Famcc

Adthqth

i

s

M

Uethtiintonnwpswesspq

WdwOpS

We used a witmaps (four levAll the participorder of increas

nstructions an

The study consby constructionDuring familiamanipulate thnstructions to

could practice made of five po

After the familof the four mapype: what are

of <name of a

Figure 1c) and answering. Dumore than thconsidered as acould continue.

After each codifficulty of thehe session, the

questionnaires,hink that cons

t? They were system.

Measures and

Usability waseffectiveness anhree measuresime to placenstruction untio answer each

number of mapnumber of corwas measuredparticipants’ cosuccessive poswell as the occexploration, wselected elemenselected. We tparticipants = questions * 8 p

We used the Shdata followed were normal, Otherwise we performed withSamples with a

thin-subjects dvels of complepants had to cosing difficulty

nd tasks

sisted in one fn and exploratiarization, pare Sucker paconstruct a maas many time

oints.

liarization phaps, and then anthe names of th

a line>? Theyselect the app

uring constructhree minutes a failure, and w.

onstruction, pe task on a 7-pey answered a S

as well as thstructing the m

also invited to

statistical me

s evaluated nd satisfaction

s: 1) time neede one object, il the object wa

h question. Effeps each participrrect answers d using the omments. Durinitions of each

currence of thewe logged for

nts as well astherefore colle288 trials for articipants = 9

hapiro-Wilk tea normal diswe computedused a Friedm

h the Wilcoxona Benjamini and

design with theexity) as indeponstruct the m(M6, M8, M10

familiarizationion tasks with tticipants wereds, how to

ap, and how to s as they want

ase, the particinswered three he two points a

y had to explopropriate line ation, when paon one instrwe provided h

participants haoint Likert scaSUS [3] and a

he following qmap helped yo

o provide any

ethods

by measurn. Efficiency wded to build th

from the fias correctly plaectiveness waspant successfuto the questioSUS questionng the construc

h Sucker pad be different instreach question

s the time at wected (6 + 8 +construction, 6 trials for exp

st to determinestribution. Whd a Univariateman test. Postn Signed-Rankd Hochberg co

e Complexity pendent variablmaps in the sam0 and M12).

n phase followthe Sucker pade told how interpret audexplore it. Thted to on a ma

ipants built eacquestions of that the extremiti

re the map (sand points befoarticipants speruction, it whelp so that th

ad to rate thale. At the end

NASA-TLX [question: do yo

ou to understan

comment on th

ing efficiencwas assessed wihe entire map;rst constructio

aced, and 3) tims assessed by thully built and thons. Satisfactionnaire and thction, we loggbeing placed, ructions. Durin

n the ID of thwhich they we+ 10 + 12) *and 4 maps *

ploration.

e if the collecthen distributioe ANOVA tet-hoc tests wes Test for Pair

orrection.

of le.

me

ed ds. to

dio ey ap

ch his ies

ee ore ent

was ey

he of

[5] ou

nd

he

cy, ith 2) on me he he on he ed as ng he

ere 8

3

ed ns st.

ere ed

Result

Constr

Twentycorrespcorrectbecaus(M6, Mbecausexperimerrors one. Fowas plpad. BattacheM12) SuckerP3 (Mpad to to 8” difficuon M1SuckerunchanSuckerhear thother S

Figure(not inexperim

of MapPost-hocompleM12 (pM10 (p

Figure

The avaveragsignificM10 aof Ma( ²=20significconditi

ts

ruction

y seven mapponded to 28tly positioned se one line didM12), and th

se the particimenter (M8, was very low or Participant 2layed before th

By the time, thed the new Supulled out a

r pad. He then M12) spent seve

the point 9, w was repeat

ulties focusing 0, he moved thr pad, so thatnged. When cr pad to attach he feedback (“aSucker pads.

7 shows the ancluding 3 mamenter). A Fri

p Complexity ooc pairwise etion time diffep<.05), betweep< .05).

e 7: Average co

map

verage time toge. Post-hoc cant difference

and M12 (p<.0ap Complexity0.85, p<.001). cant completiions (p<.05) as

ps out of 3283 Sucker pad

and linked. Td not start frohree maps weipants requireM12, M12). (5) we provid

2 (P2) in M6, the user effectivhe user had moucker pad to astring too strreattached it t

eral minutes trwhereas the instted several ton the task: w

he extremity oat the guidancconstructing M

a new one, buattached”), and

average time taps constructeiedman test sh

on completion comparison

ferences betweeen M12 and M

ompletion times

ps (error bars ar

o place a Tanpairwise co

e in time neede05). There wasy on rough guPost-hoc pairwon time diffes shown in Figu

2 were correds out of 28

Two maps werom the right Sere considereded assistance Because the n

de further detaithe instruction vely attached oved away andan incorrect onrongly and deto the wrong Srying to attachtruction “attachtimes. P8 ewhen placing f the string ins

ce instructionsM12, he foundut did not waitd then tried to

to construct coed with the howed a signifi

times ( ²=16.4n revealed en M6 and M8

M8, and between

s (in seconds) to

re IC95).

ngible Reel waomparison reed to place a Ts also a signifiuidance timewise comparisoerences betweure 8.

ect, which 8 (98.3%)

re incorrect Sucker pad d incorrect

from the number of ils on each “attached” the Sucker d therefore ne. P1 (on

etached the Sucker pad. h a Sucker h an object xperienced one object

stead of the s remained d the good t enough to attach it to

orrect maps help of the icant effect

4, p<.001). significant

8, M10 and n M12 and

o construct

as 23.7 on evealed a

TR between icant effect per object

on revealed en all the

F

t

E

Tn(

MppctoAepef

(

TqM1thcampli

Q

TPp9rin

Figure 8. Avera

to correctly pla

to a previous on

following roug

fin

Exploration

Table 2 indicanumber of po(correct answe

Map Complexip=.01). Post-hparticipants selcompared to Mo answer each

ANOVA with exploration timp<.01). A poexploration timfor M8 (p<.05)

M6

(a) 14.6 (11.

(b) 3.9 (0.9

Table 2. Avera

and averag

The percentagquestions wereM10, and 79.2%1) participants he wrong line

constructed (P1a misalignmenmap (P8 twiceperform the poines (P3, P6).

Questionnaires

The average Participants hapoint Likert sca9 illustrates therather easy (1, 2ndicates the sc

age times and IC

ace a Tangible R

ne or placing it

gh guidance ins

ne guidance inst

ates the averagoints and lineers only). We

ty and Numbehoc pairwiselected less ele

M8, M10 and M question follo

Map Complmes differed

ost-hoc Tukeymes for M12 w and M6 (p<.0

M8 1) 22.5 (17.7

9) 6.0 (1.0)

age time to answ

ge number of ele

indicated in

ges of correcte: 91.7% for M% for M12. Erpointed to the (P2 twice, P3,

1, P2); 3) Sucknt between thee); 4) participointing gesture

s

SUS score ad to rate the ale (1 = very ee number of pa2), normal (3, 4

cores of the NA

C95 for the thre

Reel: 1) attachi

anywhere on th

structions (gree

tructions (oran

ge completions selected forfound an inte

r of elements se comparisonements when eM12 (p<.05). Cowed a normal lexity as fact

significantly y’s HSD test were significan1).

M10 7) 31.4 (23.9

) 7.3 (2.0)

wer one explora

ements selected

n parenthesis.

t answers to M6; 95.8% forrrors were dueright intersecti

, P6); 2) maps ker pads got de

tangible mappants had troues and therefo

for the systdifficulty of t

easy; 7 = very darticipants who4, 5) or difficu

ASA-TLX ques

ee steps require

ng a Sucker pa

he table (blue);

n); 3) following

ge).

n times, and thr each questioeraction betwe

selected ( ²= 9n showed thexploring M6

Completion timdistribution. A

or showed th(F(3,28)=5.7indicated th

ntly higher th

M12 ) 41.9 (27.8)

8.6 (0.8)

tion question (a

d (b). SDs are

the exploratior M8; 91.7% fe to the fact thaion but followwere incorrect

etached, creatinp and the digituble to correctore to select th

tem was 83.the task on a difficult). Figuo found the ta

ult (6, 7). Tablestionnaire.

ed

d

2)

g

he on en

91, hat as

mes An hat 70, hat an

)

a)

on for at: ed tly ng tal tly he

.6. 7-

ure ask e 3

Figure

task

Conclu

Constr

Resultsconstruissues. most cminutequicklyprovedto follsuggesby imparticipstated tto makP2, P3the expwas nefailed “couldobservbecaus

Explora

89.6% systemHowevdifficuonly, aand pospent oP7, whsecondintersemanag

Men

Phys

Temp

Pe

F

Ta

partic

Nu

mb

er

of

pa

rtc

ipa

nts

e 9: for each ma

rather easy (bl

usions about

ruction

s shows that uct the maps 85% of the momplex map w

es. The rough gy move a Sucd to be efficienlow fine thansts that the ove

mproving the pants found ththat “instructio

ke mistakes” an, P6. Howeverperiment they ecessary to rein constructi

dn’t hear the ived that P4 fase he did not ke

ration

of the answem is usable to ver, we observulties to perforand did not alwoints. For examon average 8.7 ho perfectly unds per elementections of twoge to select the

ntal demand

sical demand

poral demand

Effort

erformance

Frustration

able 3. For each

cipants who ind

ap, number of p

lue); normal (gr

map construc

most of the without exp

maps were correwas constructedguidance techncker pad close nt. On average n rough guidaerall completio

fine guidanche instructionsons are extremnd similar comr, P3 and P5 dwere getting

emain concenting M12 becinstructions an

failed in consteep on focusing

ers were correc retrieve speced two issues: rm pointing g

ways manage tmple, P3 kept u

seconds to selnderstood the t. 2) Some pao lines and thline they want

<33

62.5

75

75

75

0

75

h NASA-TLX c

dicated a value i

participants who

reen); difficult (

ction and expl

participants mperiencing anyectly constructd in approximanique allowed t

to the target it was three tim

ance instructioon time could bce mode. Mos easy to unde

mely clear, it is mments were m

eclared that at tired, notably trated (P3 sai

cause she feltnymore”). Simtructing M10 g on the instru

ct, which showcific spatial in1) some partic

gestures with to quickly selecusing several flect one elemengesture, spent

articipants poinherefore did nted to.

<66 <

25 1

0

25

25

12.5 8

25

criteria, percent

inferior to 33, 6

o found the

(orange).

loration

managed to y usability ed, and the

atively four the users to and hence

mes longer ons, which be reduced

ost of the erstand. P4 impossible

made by P1, the end of because it

d that she t like she

milarly, we and M12

uctions.

ws that the nformation. cipants had one finger ct the lines fingers and nt, whereas t only 0.96 nted at the not always

<100

12.5

25

0

0

87.5

0

tage of

66 and 100.

Visual Impairment and Technology #chi4good, CHI 2016, San Jose, CA, USA

C

FthtwastoRdicinuwrcth

Obhp“liwrppth(ua

S

Opointhawmbo

D

Omtoliliebth((esim

Complexity

Four participanhe last three cwo conditions

at reconstructinstated that it beogether when

Reels. Two subdifficult to unds interesting to

completion timncrease with t

usability issueswere likely dueresult from thconditions happhe complexity

Overall, we dbetween the horizontal lineplace an objec“new object” iinked to a pre

was being conrevealed a poprobably reflecplace a new Suhe maps that c

(number and suser’s ability tappropriate acti

Satisfaction and

Overall, particperformance (Nof the participndicated by thhey were able

and two pinpowere explicitlymaps. Finally, be very helpfulorientation and

DISCUSSION

Overall, we shmaterialize poio move and stink two or mines. The pre-

easily manipulbuilt with themhat maps of d

(85%) and can(89.6% of correfficient (23.7ssatisfying (SUSmproved by a

nts found that onditions (P1,(P4 and P7). T

ng M12 foundecame difficultthey were alr

bjects (P4 and derstand and exo note that excemes to place onthe map comps. Three failure to a lack of che length of pened at the enof the map per

did not obsertimes to cos. However, w

ct anywhere oinstruction (i.eevious one) grnstructed (a Positive correlatcts the time nucker pad, sugcan be built is size of Tangibto interpret auions.

d usefulness

cipants were NASA-TLX scpants), and fe SUS score. S

e to understandinted that it wy asked to unthree participal for educationmobility lesso

howed that Tants and lines otable. A reel w

more phicons tstudy showed late Tangible m. Results of different compn be exploredrect answers).s only in aveS of 83.6). Th

allowing multip

difficulty rem P2, P4, and P

The two particid this task diffit to attach twoready close toP7) declared thxplore more coept between Mne object did nplexity, neitherres (in M10, concentration tthe whole exnd of the exper se.

rve any statisnstruct obliquwe observed ton the table ae. an object thradually increa

Pearson correlation : r=0.64,

needed to find ggesting that thlimited by phyble Reels) rathudio instructio

highly satiscore superior tfound the sysSix participantd the maps aft

would have beenderstand and

ants said that thnal purposes (mons).

angible Reels on tangible mawith strong matogether and pthat participanReels, and uthe main stud

plexity are ved by the users In addition,

erage to place he exploration ple fingers sel

mained similarP5) or in the lapants who fail

ficult. P1 and Po Tangible Reeo other Tangibhat it will be toomplex maps.

M10 and M12, thnot significantr the number M12 and M1that could eith

xperiment (theeriment) or fro

stical differenue, vertical that the time

after hearing thhat must not bases as the maation coefficie, p<0.01). Throom where

he complexity ysical constrainher than by th

ons and perfor

sfied with thto 66% for 85stem usable, s mentioned th

fter constructioen easier if thd memorize thhe system woumathematics an

are efficient aps, and are eaagnets is used provide tangibnts were able

understand mady demonstratry often corres with accurathe system wan object) anmode could b

lection of poin

in ast ed P5 els ble oo It

he tly of 2)

her ese om

nce or to he be ap

ent his to of

nts he rm

he 5% as

hat on, ey he

uld nd

to asy

to ble to

aps ed ect cy

was nd be nts

and licrossinWeighheight are use

Map co

The nutable limost ocomplewhen oamounmateriaReels, using feedba

Design

We detangiblthroughtexture(calledbend tprovidparticuexampmode, modifireposit

More t

With tpossiblpotentiincludicharts, graphicmay ac

CONC

Graphiused itechnothem atangiblexplorithe desbe usethey aconvey

ines, and prongs. Howeverhts hindered ma

of phicons shoed by visually i

omplexity

umber of Tangimits the compof the participex maps, somobjects were t

nt of informatioalize the mosand then provgestural inter

ack (see [22] fo

ning advanced

esigned a solule. Areas coulh audio (see

es [6]. Besided “modifiers”) them to constre advanced

ular points of iple, for zoomin

and move apaied accordingltion Tangible R

than tangible m

the materializale to constructially materializing points, line etc.). Figurcal representaccess with Tan

Figure 10. Ot

LUSION

ics (maps, orgin education

ologies that heare still uncomle tabletop iing tangible gsign of Tangibled to materialiare stable, easyy spatial repres

oviding specir, several parap explorationould be carefulimpaired users

gible Reels thatplexity of the mpants managed

me reported thatoo close to eaon is required, st important evide access to ractions as weor instance).

d functions

ution to makeld also be rep[27] for exam

s, we started that are atta

ruct curves. Tfunctions

interests, or zong, users will art two objectsly and the uReels and place

maps, tangibl

ation of pointst maps. Obviouze any type ofes, and areas

re 10 illustraations that a vngible Reels.

her graphics wi

ganigrams, barbut also in elp visually immmon. In thisinterface that

graphics. Morele Reels, a newize points andy to manipulasentations of di

ific feedback rticipants rep

n, which suggelly considered

s.

t can be placedmap. In additiond to constructat the task waach other. Wheit would be int

elements usingless importan

ell as audio a

e both points presented andmple) or illusto design sm

ached to the sTangible Reels

like displayooming and pahave to select. The digital m

user will be e new ones.

e graphics

s and lines, wusly, Tangible

f graphical rep(graphs, flow

ates three exavisually impair

ith Tangible Re

r charts, etc.)everyday life.

mpaired users ts paper we intt allows buie precisely, wew type of phicod lines. We shate, and can bifferent comple

for lines ported that sts that the when they

d above the n, although t the more as difficult en a larger teresting to g Tangible nt elements and haptic

and lines d conveyed sory tactile

mall objects strings and s may also ying/hiding anning. For t the zoom

map will be guided to

we made it e Reels can resentation charts, bar amples of red person

eels

are widely . However to perceive troduced a lding and

e described ons that can howed that be used to exity.

ACKNOWLEDGMENTS

We thank the LACII and notably Claude Griet, Laurence Boulade and Nathalie Bedouin. We are very grateful to all the participants of the study. We also thank Gilles Bailly, Anke Brock and Bernard Oriola for their meaningful comments. This work is part of the AccessiMap project (research grant AccessiMap ANR-14-CE17-0018).

REFERENCES

1. G Blasko, Chandra Narayanaswami, and Steven Feiner. 2006. Prototyping retractable string-based interaction techniques for dual-display mobile devices. Proceedings of the SIGCHI Conference on

Human Factors in Computing Systems (CHI ’06): 369–372. http://doi.org/10.1145/1124772.1124827

2. Anke M. Brock, Philippe Truillet, Bernard Oriola, Delphine Picard, and Christophe Jouffrais. 2015. Interactivity Improves Usability of Geographic Maps for Visually Impaired People. Human-

Computer Interaction 30: 156–194.

3. John Brooke. 1996. SUS: A “quick and dirty” usability scale. In Usability Evaluation in Industry, P. W. Jordan, B. Thomas, B. A. Weerdmeester and I. L. McClelland (eds.). Taylor & Francis, London, UK, 189–194.

4. Polly Edman. 1992. Tactile graphics. AFB press, New York, USA.

5. Sandra G. Hart and Lowell E. Staveland. 1988. Development of NASA-TLX (Task Load Index): Results of Empirical and Theoretical Research. In Human Mental Workload, Peter A. Hancock and Najmedin Meshkati (eds.). Elsevier, 139–183. http://doi.org/10.1016/S0166-4115(08)62386-9

6. V. Hayward. 2008. A brief taxonomy of tactile illusions and demonstrations that can be done in a hardware store. Brain research bulletin 75, 6: 742–752.

7. M. T Horn. TopCode: Tangible Object Placement Codes. Retrieved from http://hci.cs.tufts.edu/topcodes/

8. CJ Huang, Ellen Yi-luen Do, and D Gross. 2003. MouseHaus Table: a Physical Interface for Urban Design. 16th Annual ACM Symposium on User

Interface Software and Technology (UIST): 41–42.

9. Hiroshi Ishii and Brygg Ullmer. 1997. Tangible bits: towards seamless interfaces between people, bits and atoms. Proceedings of the ACM SIGCHI

Conference on Human factors in computing

systems: 234–241. http://doi.org/10.1145/258549.258715

10. R.D. Jacobson. 1998. Navigating maps with little or no sight: An audio-tactile approach. Proceedings of

Content Visualization and Intermedia

Representations, 95–102. Retrieved from http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.21.2860

11. Nikolaos Kaklanis, Konstantinos Votis, Panagiotis Moschonas, and Dimitrios Tzovaras. 2011. HapticRiaMaps: Towards Interactive exploration of web world maps for the visually impaired. Proceedings of the International Cross-

Disciplinary Conference on Web Accessibility -

W4A ’11, ACM Press, 20.

12. T. Kaltenbrunner, M., Bovermann, R. Bencina, and E Costanza. 2005. TUIO - A Protocol for Table-Top Tangible User Interfaces. Proceedings of the

6th International Workshop on Gesture in Human-

Computer Interaction and Simulation (GW 2005).

13. Shaun K. Kane, Meredith Ringel Morris, Annuska Z. Perkins, Daniel Wigdor, Richard E. Ladner, and Jacob O. Wobbrock. 2011. Access Overlays: Improving Non-Visual Access to Large Touch Screens for Blind Users. Proceedings of the 24th

annual ACM symposium on User interface software

and technology - UIST ’11, ACM Press, 273–282. http://doi.org/10.1145/2047196.2047232

14. Erik Koch and Hendrik Witt. 2008. Prototyping a chest-worn string-based wearable input device. 2008 International Symposium on a World of

Wireless, Mobile and Multimedia Networks: 1–6. http://doi.org/10.1109/WOWMOM.2008.4594882

15. Uwe Laufs, Christopher Ruff, and Jan Zibuschka. 2010. MT4j - A Cross-platform Multi-touch Development Framework. ACM EICS 2010,

Workshop: Engineering patterns for multi-touch

interfaces, ACM, 52–57.

16. AK Lobben. 2005. Identifying the needs of tactile map makers. Proceedings of XXII International

Cartographic Conference.

17. Muhanad S Manshad, Enrico Pontelli, and Shakir J. Manshad. 2012. Trackable Interactive Multimodal Manipulatives: Towards a Tangible User Environment for the Blind. Proceedings of ICCHP

2012, Springer Berlin Heidelberg, 664–671. http://doi.org/10.1007/978-3-642-31534-3

18. David McGookin, Euan Robertson, and Stephen Brewster. 2010. Clutching at Straws: Using Tangible Interaction to Provide Non-Visual Access to Graphs. Proceedings of the 28th international

conference on Human factors in computing systems

- CHI ’10, ACM Press, 1715–1724. http://doi.org/10.1145/1753326.1753583

19. Joshua A. Miele, Steven Landau, and Deborah Gilden. 2006. Talking TMAP: Automated generation of audio-tactile maps using Smith-

Kettlewell’s TMAP software. British Journal of

Visual Impairment 24, 2: 93–100. http://doi.org/10.1177/0264619606064436

20. Peter Parente and Gary Bishop. 2003. BATS : The Blind Audio Tactile Mapping System. Proceedings

of ACM South Eastern Conference, ACM Press.

21. Norman Pohl, Steve Hodges, John Helmes, Nicolas Villar, and Tim Paek. 2013. An interactive belt-worn badge with a retractable string-based input mechanism. Proceedings of the SIGCHI

Conference on Human Factors in Computing

Systems - CHI ’13: 1465. http://doi.org/10.1145/2470654.2466194

22. Benjamin Poppinga, Charlotte Magnusson, Martin Pielot, and Kirsten Rassmus-Gröhn. 2011. TouchOver map: Audio-Tactile Exploration of Interactive Maps. Proceedings of the 13th

International Conference on Human Computer

Interaction with Mobile Devices and Services -

MobileHCI ’11, ACM Press, 545–550. http://doi.org/10.1145/2037373.2037458

23. Bernhard Schmitz and Thomas Ertl. 2012. Interactively Displaying Maps on a Tactile Graphics Display. SKALID 2012–Spatial

Knowledge Acquisition with Limited Information

Displays (2012), 13–18.

24. Jochen Schneider and Thomas Strothotte. 2000. Constructive exploration of Spatial Information by Blind Users. Proceedings of the fourth

international ACM conference on Assistive

technologies: 188–192. http://doi.org/10.1145/354324.354375

25. Ehud Sharlin, Benjamin Watson, Yoshifumi Kitamura, et al. 2004. The Tangible Pathfinder Design of a Wayfinding Trainer for the Visually Impaired. Proc. Graphics Interface: 2–3.

26. Mathieu Simonnet, Dan Jacobson, Stephane Vieilledent, and Jacques Tisseau. 2009. SeaTouch: a haptic and auditory maritime environment for non visual cognitive mapping of blind sailors. COSIT

2009, LNCS 5756, Springer-Verlag, 212–226. http://doi.org/10.1007/978-3-642-03832-7_13

27. The Braille Authority of North America. 2010. Guidelines and Standards for Tactile Graphics. Retrieved from http://brailleauthority.org/tg/web-manual/index.html

28. J Underkoffler and H Ishii. 1999. Urp: a luminous-tangible workbench for urban planning and design. Proceedings of the SIGCHI conference on Human

Factors in Computing Systems: 386–383. http://doi.org/10.1145/302979.303114

29. Fernando Vidal-verdú and Moustapha Hafez. 2007. Graphical Tactile Displays for Visually-Impaired People. Neural Systems and Rehabilitation

Engineering, IEEE Transactions on 15, 1: 119–130.

30. Zheshen Wang, Baoxin Li, Terri Hedgpeth, and Teresa Haven. 2009. Instant Tactile-Audio Map: Enabling Access to Digital Maps for People with Visual Impairment. Proceeding of the eleventh

international ACM SIGACCESS conference on

Computers and accessibility - ASSETS ’09, ACM Press, 43–50. http://doi.org/10.1145/1639642.1639652

31. Limin Zeng and Gerhard Weber. 2011. Accessible Maps for the Visually Impaired. Proc. of IFIP

INTERACT 2011 Workshop on ADDW, 54–60.

32. Limin Zeng and Gerhard Weber. 2012. ATMap: Annotated Tactile Maps for the Visually Impaired. COST 2102 International Training School,

Cognitive Behavioural Systems, LNCS Volume

7403, 2012, Springer Berlin Heidelberg, 290–298. http://doi.org/10.1007/978-3-642-34584-5

33. Haixia Zhao, Catherine Plaisant, Ben Shneiderman, and Jonathan Lazar. 2008. Data Sonification for Users with Visual Impairment. ACM Transactions

on Computer-Human Interaction 15, 1: 1–28.