Future Alternatives to User Testing: Virtual User Modelling in Inclusive Product Design

tripastroturfAI and Robotics

Nov 7, 2013 (3 years and 7 months ago)

87 views

Future Alternatives to User Testing: Virtual User
Modelling in Inclusive Product Design


Presenter: Joshue O Connor


Authors:

Joshue O Connor, Antoinette Fennell, Patrick Klein, Markus Modzelewski,
Michael Lawo, Yehya Mohamad, Pierre Kirisci, Thomas Fiddia
n, Christopher
Bowden.


Keywords:

Universal Design, User Testing, Assistive Technology, Product
Design, Virtual User Modelling, eInclusion.


ABSTRACT


Traditionally, the needs of people with physical impairments such as visual,
hearing, and dexterity impai
rments are often just not sufficiently considered by
industry when designing user interfaces for consumer products. This is
exacerbated by the fact that it is not unusual for an individual to have multiple
impairments, for example when elderly people may e
xperience hearing and
sight loss as well as loss of dexterity.


Now the terms ‘inclusive design’, and ‘universal design’ are becoming more
common within the designers vernacular. There is a greater awareness of the
value of inclusive design methodologies
for both designer and end user, such
as the user testing of product prototypes.


However, from the perspective of industry there are some distinct barriers that
have impeded the adoption of inclusive design such as a perceived lack of a
market or confusio
n as to how best approach developing inclusive design
methodologies. Part of the aim of Universal Design is to bring accessible and
usable design for everybody out from its niche and mainstream it, but how will
this be realistically achieved?


In practical

terms, Universal Design methodologies must ideally compliment
the existing product design workflows of designer. Or at the very least be as
disruptive as possible. Ideally, they should enhance how designers currently
do things, and put the least cognitive

load as possible on the designer. So
ideally tools, that support the tenants of Universal Design, that are not difficult
to use and that can plug into common existing tools are desirable.



Designing consumer products that are truly inclusive presents man
y varied
and interesting challenges for the designer and User testing is becoming more
established as a valuable part of the inclusive design process. However, the
costs involved in doing user testing can be substantial, in terms of time, as
well as financ
ially. So are there alternatives?


Virtual User Modelling (VUM) extends traditional user modelling and involves
the use of the 3D or virtual environment throughout the entire product
development process. Virtual User Profiles (a customisable library of vir
tual
people) are developed to provide simulations of real world interaction
paradigms. Even though the use of VUM is currently relatively rare they have
previously being used in industries as diverse as the automobile, fire fighting
industries and now the
area of social inclusion and Universal Design are
getting attention through the EU FP7 project cluster Virtual User Modelling
and Simulation Cluster (VUMS).


As a part of this cluster, that aims to look at mainstreaming accessibility and
improving support

and tools for designers, the VICON project aims to explore
the potential of the VUM, and to find out to what extent can the use of a VUM
compliment or replace traditional user testing?


The VICON Project

comprises three “How
-
To” questions:



How can consu
mer
-
oriented companies adapt their existing product portfolios
to the needs of sensory and physically impaired users in a competitive and
sustainable manner?


How can companies in Europe gain a competitive advantage from adopting
inclusive design, related
to markets characterized by cheap products from Far
East?


How can product developers be empowered during product development to
consider the real
-
time accessibility needs of physically impaired users, with
the aim of securing universal access to future co
nsumer products?


This paper/presentation will give an overview of the VICON project
[http://vicon
-
project.eu/] and examine the potential role that Virtual User
Modelling (VUM) has to play as a part of inclusive product design, as well as
looking at some o
f the technical aspects of developing a VUM.


The Business Case


Generally, the needs of people with physical impairments such as visual,
hearing, and dexterity impairments are just not sufficiently considered by
industry when designing user interfaces fo
r consumer products. This is
exacerbated by the fact that it is not uncommon for an individual to have
multiple impairments, which are particularly common among elderly people.


What industry seems to be gradually realising is that both people with
disabi
lities and older people have money to spend. This demographic
therefore represent a large potential market that can be tapped into. This is a
substantial driver in the current surge in industrial interest in Universal Design,
and it is welcome.


From the p
erspective of industry there are some distinct issues that have
impeded the adoption of inclusive design. They are (in no particular order of
importance):


1)

The cost associated with designing niche products that will be used by
a minority group, such as the

market for Assistive Technology devices
that it relatively small compared to the mainstream market.

2)

Perceived lack of a market.

3)

The high cost of user centered design processes such as iterative user
testing, focus groups, and expert evaluation. For a prod
uct that is to be
truly inclusive (from its design inception) these costs can aggregate
easily.

4)

Amongst designers, there has been a general dearth of knowledge as
to exactly how to correctly design inclusive products that can also be
used by people with di
sabilities. This has lead to the disconnect
between designing for people with disabilities and the average
consumer
-

creating a kind of product segregation. Universal Design
(UD) aims to bridge this divide and show designers that the two
domains are not o
rthogonal.


It is also worth noting that for the consumer with disabilities the lack of
awareness of Universal Design has meant:


1)

The high cost to the consumer of products that can be considered
Assistive Technology.

2)

There are many examples of technologie
s that are in by design
‘inclusive’ but end up being niche


and very expensive. Penny and
Giles joysticks, or DAISY players for example. They are also, while
useful to the groups that they serve, often not aesthetically pleasing
and can look ‘remedial’ la
ck or style at worst, or at best making the
user self conscious or uncomfortable.


There have been some very high profile advances in technology that includes
Assistive Technology functionality embedded into the system without the
product distinguishing it
self as ‘different’, such as the iPhone and its built in
Voice Over TTS software which has resulted in the increasing popularity of
the device amongst blind and vision impaired users. [1]


There are other far more low
-
tech examples of Universal Design, whe
re the
UD component is invisible and the product is just easier to use. For example,
the perforated
Panasonic battery pack means the batteries are easy to
remove
, or the Oxo Good Grips range of kitchen utensils. [2] [3]


The Challenge of Inclusive Design


Designing consumer products that are truly inclusive presents many varied
and interesting challenges for the designer. Part of the aim of Universal
Design is to bring accessible and usable design out from its niche and
mainstream it. When Universal Desig
n truly becomes ubiquitous then
inclusive thinking shall be the default setting of the designers make up


where the needs of people with disabilities and older people reflect just
another aspect of humanities diversity which must be accommodated in the
pr
oducts that they build.



These are challenges that need to be enthusiastically embraced by the
designer. The design of inclusive products has the potential to not only enrich
the user but the designer also. When successful, inclusive design can be at
the
heart of many innovative breakthroughs. At a time when consumer
confidence is low, and fiscal concerns are foremost in everyone’s mind the
manufacture and use of substandard unusable products needs to be
minimised. There is no reason why Universal Design c
annot be at the heart of
a new wave of innovation in mainstream product design that matches not only
the consumers’ expectation but also their ability.


However, it is only when mainstream product manufactures take up cause of
Universal Design that there i
s a chance of this becoming a reality. So how can
this happen? In practical terms, Universal Design methodologies must ideally
compliment their existing product design workflows in the least disruptive
ways possible. Ideally, they will enhance how designer
s currently do things,
and put the least cognitive load as possible on the designer. So ideally tools,
that support the tenants of Universal Design, that are not difficult to use and
that can plug into common existing tools are desirable.




What is Virtua
l User Modelling?


Traditional User modelling is a way of doing

predictive evaluation of real
-
world
tasks
. It is achieved

by trying to represent some aspect of the user's
understanding, knowledge, intentions or
mental processes.
User models can
be divided
into the following three categories:





Hierarchical representation of the user's tasks and goal structure




Linguistic and grammatical models




Physical and device level models





The first category
covers the

formulation of goals and tasks.

The second ca
tegory
covers the grammar of how articulation translation
impacts the mental model, and therefore
how
a product or interface can be

understood by the user.

The third category deals with articulation at the human motor level.




GOMS and Fitt's Law are

two

widely adopted

techniques in user modelling,
and applying these kind of user modelling techniques to the virtual
environments that product designers use today is called Virtual User
Modelling (VUM). [4] [5] [6]



Virtual User Modelling (VUM) extends t
raditional user modelling and involves
the use of the 3D or virtual environment
throughout the entire
product
development process. Virtual User Profiles (a customizable library of virtual
people)
are developed to provide simulations of real world interacti
on
paradigms. Even though the use of VUM is currently relatively rare they have
previously being used in industries as diverse as the automobile, fire fighting
industries and now the area of social inclusion and Universal Design are
getting attention throu
gh the FP7 project cluster
Virtual User Modelling and
Simulation Cluster

(VUMS). [7]



eInclusion and the VUM Cluster


Sophisticated Virtual User Modelling techniques have the potential to be used
as a part of the inclusive design toolkit. Projects like VI
CON explicitly aim to
explore the validity of using Virtual User Modelling as a tool for the creation of
inclusively designed products. [8]


The cluster of projects includes:


The VERITAS Project

aims to develop, validate and assess tools for built
-
in
acce
ssibility support at all stages of ICT and non
-
ICT product development,
including specification, design, development and testing.

VERITAS will
introduce simulation

based and virtual re
ality testing at all stages of Assistive
T
echnologies product design and

development into the automotive, smart
living spaces (buildings & construction, domotics), workplace, ehealth and
infotainment applications areas.
[9]


The MyUI Project

will foster t
he mainstreaming of accessible
ICT products


a major issue for e
-
Inclusi
on. The project addresses important barriers that
include developers’ lack of awareness and expertise, time and cost
requirements of incorporating accessibility and missing validated approaches
and infrastructur
es.


The project’s approach extends
the notio
n of Universal
Design by addressing specific user n
eeds through adaptive personalised
interfaces. [10]


The GUIDE project

aims to fulfil

the individual accessibility needs of elderly
users with mild impairments through the provision of a development toolbo
x
for personalized, adaptive multimodal user interfaces. This GUIDE toolbox will
especially put developers of TV set
-
top box applications in the position to
realize applications with accessibility features tailored to the specific needs of
impaired users w
ith reduced development risk and costs.
[11]



The VICON Project
comprises three “How
-
To” questions:





How can consumer
-
oriented companies adapt their existing product
portfolios to the needs of sensory and physically impaired users in a
competitive and su
stainable manner?




How can companies in Europe gain a competitive advantage from
adopting inclusive design, related to markets characterized by cheap
products from Far East?




How can product developers be empowered during product development
to consider th
e real
-
time accessibility needs of physically impaired users,
with the aim of securing universal acc
ess to future consumer products? [8]


The Validity of the VUM in real world product design
and modelling humans as Avatars


For the modelling of virtual use
rs as avatars there is a widely used standard of
the H
-
Anim group (ISO/IEC 19774), it specifies a system for representing
humans in a 3D environment. Conceptually, each humanoid is an articulated
character (that is really just a complex data structure) tha
t can be embedded
in different applications across multiple platforms and animated. [12]


Simulation is another way of testing by building Virtual Reality (VR)
environments with both the product and the avatar, and let user
representatives evaluate the con
cepts by watching the VR simulation and
interacting with it. This method can be used as away to generate early user
feedback of system prototypes, with can improve the product design is used
as a part of an iterative design process.


The adoption of

Virtua
l User Models

has the potential to be incorporated into
existing product design workflows by product design teams, and providing a
mean of giving feedback on the needs of various types of end users in the
early design phase, before more expensive ‘hard’ p
rototypes are built.


The idea is that if accessibility and usability problems can be found early and
eradicated using existing Computer
-
Aided Design (CAD) then the previously
mentioned costs that have been associated with User Centered Design
processes ca
n be reduced. The aim of the VICON project is to test the validity
of this model.


So how will it work?


The following are some technical details giving an overview of how a VUM
may be implemented as a part of the VICON project. Please note the
following i
s not normative and is subject to change.






Ontology
-
based Reasoners,
Sem
antic Reasoners and
User Profiling Ontology


The Virtual User Model itself will be an ontology
-
based Reasoner, where
databases are in one of the semantic web standards format. Cu
rrently we are
testing some formats and how we can model our User.


Semantic Reasoner/User Profiling Ontology


A reasoning engine, rules engine, or simply a reasoner, is a piece of software
able to infer logical consequences from a set of asserted fact
s or axioms. The
notion
of a semantic reasoner generalis
es that of an inference engine, by
providing a richer set of mechanisms to work with.


The inference rules are commonly specified by means of an ontology
language, and often a description language.
Use of probabilist
ic reasoners,
Bayesian networks
etc

could potentially also be used.



User Profiling Ontolog
ies

with support for situation dependent preferences
(which could be the user environment)

and c
ontext
-
aware adaptive systems
are also

useful.


Practical
Mappings between Ontologies


There are potential mappings between
a new VCard ontology and other
vocabularies used to describe people such as W3C's PIM vocabulary and the
FOAF vocabulary.
Information from our user trials will be used as a basi
s for
design
ing the user and
environment model
s
.
[13]


Formats for User Definition


What follow are samples of potential ways of defining qualities of the user and
their environment. Note these examples are illustrative in a very broad sense;
the final mo
dels will have a much greater level of granularity. The user model
outlined here is an example from the VICON Virtual User Model Draft. [14]


For modelling the user themselves:


<?xml version="1.0"?> <rdf:RDF
xmlns:rdf="http://www.w3.org/1999/02/22
-
rdf
-
syn
tax
-
ns#"


xmlns:owl="http://www.w3.org/2002/07/owl#"


xmlns:Vicon=

http://www.vicon
-
project.eu/ontologies/Vicon.owl#



<rdf:Description rdf:about="http://www.vicon
-
project.eu/ontologie
s/Vicon.owl#P5">


<rdf:type rdf:resource="http://www.vicon
-
project.eu/ontologies/Vicon.owl#UserModel"/>


<Vicon:UserModelName>Gandalf</Vicon:UserModelName>


<Vicon:UserModelFieldOfVision
rdf:datatype="http://www.w3.org/2001/XMLSchema#int">2</Vicon
:UserModelFieldOfVision>


<Vicon:UserModelAge
rdf:datatype="http://www.w3.org/2001/XMLSchema#int">80</Vico
n:UserModelAge>


<Vicon:UserModelGender>M</Vicon:UserModelGender>


<Vicon:UserModelGlasses>Y</Vicon:UserModelGlasses>

</rdf:Description>

</r
df:RDF>



** RULE for 2 different visual impaired user profiles,
depending on if the user needs glasses or not:

// Visual Impairment Profiles

//VI0

equal(?glasses,"N")

-
> (?x rdf:type Vicon:VProfile0).

//VI1

equal(?glasses,"Y")

-
> (?x rdf:type Vicon:VPro
file1).



** The above example uses

ICF Profiles

that allow th
e
reasoning engine to parse the
input model components
like if
the user needs glasses to read as members of different
profile groups
into the virtual enviro
n
ment.
[15]



This could make VUM sche
ma simple to author, the reasoning engine can
handle the development of the VUM etc. In practice the designer will probably
select their desired user persona from a preset array of user types such as
low vision and other combinations of ability.

The system

itself will reason the
predefined profile based on the specified rule sets, which profile group the
designer would target with his product.


Note that there are many other ways of using RDF schema to model the user
and the final working model may take a d
ifferent approach with a greater level
of predicates, will be able to take integers, strings, and
B
oolean
values/operators as needed. Other anthropomorphic databases may be used
as references to build the VUM.



RDF Format:
Defining the Environment

Also it

is possible to describe the user environment using RDF. For example,
if during a test the following notes were made about a users environment:


NOTE:
Kitchen with background noise 9 of 10.


Here is an example of an RDF schema that describes that environme
nt.


<?xml version="1.0"?>


xmlns:owl="http://www.w3.org/2002/07/owl#"


xmlns:Vicon=”
http://www.vicon
-
project.eu/ontologies/Vicon.owl#



<rdf:Description rdf:about="http://www.vicon
-
p
roject.eu/ontologi
es/Vicon.owl#Kitchen
">


<Vicon:Env
ironmentBackground_noise

rdf:datatype="http://www.w3.org/2001/XMLSchema#int">9</Vicon
:
EnvironmentBackground_noise
>


<Vicon:EnvironmentName>Kitchen
</Vicon:EnvironmentName>



</rdf:Description>

</r
df:RDF>


RDF Format: Defining objects in the environment

The make the definition of objects as part of an environment possible, the
VICON approach creates new classes based on the environment individuals
using the same name. For example every object, tha
t is located in the kitchen
environment, consists of the following type definition:


<rdf:type rdf:resource="http://www.vicon
-
project.eu/ontologies/
Vicon.owl#Kitchen
"/>


Format for user model

Currently the User Model is still being developed, therefore t
he tools used to
define the environment are potentially in flux, but in terms of a sample of
/how/

this will be achieved the above is sound
.

Whether it is an XML VCARD model
or an RDF model for the user environment
-

that the semantic reasoner can
interact

with.


Jena: Framework for Semantic Web Apps

A possible API is Jena, which is a Java framework for building Semantic Web
applications. It provides a programmatic environment for RDF, RDFS and
OWL, SPARQL and includes a rule
-
based inference engine.
[1
6]


The Jena Framework includes:



A RDF API



Reading and writing RDF in RDF/XML,



N3 and N
-
Triples



An OWL API



In
-
memory and persistent storage



SPARQL query engine


We therefore aim to build 'templates' of user and environment information
containing ver
y specific attributes

or referencing specific attributes

that could
be useful for the desig
n of a virtual user environment
.



Some examples of Virtual Design tools

CATIA/RAMSIS RAMSIS is a 3D
-
CAD
-
ergonomics tool, designed in
collaboration with the Germa
n automobile industry for the ergonomic
development of vehicles and cockpits. Special functions are the realistic
replay of international body data, as well as efficient analysis for visibility,
comfort and ergonomics formulations. RAMSIS is a major CAD to
ol for
ergonomics design and analysis of vehicle interiors and working places and is
already used by 70% of automobile manufacturers.
[17]


Working with VUMS
and the potential of the VUM as
an alternative to User Testing?

Regardless of how the VUM is const
ructed, it is hoped that the VICON project
can provide test cases, as a proof of concept to see if the VUM can be used
to partially replace current user testing for product designers. This is because
user testing, while it is very useful, can be expensive
and time consuming.


The VICON project may also give us some idea of what the VUM will not be
able achieve which will also be very useful, and we may find that the actual
contact designers have with end users via the user testing process is in itself
valua
ble though often hard to quantify. [18]


Reference


[1]
http://www.apple.com/accessibility/voiceover/

[2]
http://universaldesign.ie/exploreampdiscover/definitionandoverview/examples/
batterypack

[3]
http://universaldesign.ie/exploreampdisc
over/definitionandoverview/examples/
oxogoodgrips

[4]
http://www.cc.gatech.edu/classes/cs6751_97_winter/Topics/user
-
model/

[5]
http://www.usabilityfirst.com/usability
-
methods/hci
-
design
-
approaches/

[6]
http://www.usabilityfirst.com/glossary/fitts
-
law/


[7]
http://www.veritas
-
project.eu/vums/

[8]
http://vicon
-
project.eu/

[9]
http://veritas
-
project.eu/2010/03
/veritas
-
press
-
release/

[10]
http://www.myui.eu/

[11]
http://www.guide
-
project.eu/

[12]
http:
//h
-
anim.org/Specifications/H
-
Anim200x/ISO_IEC_FCD_19774/

[13]
http://www.ibiblio.org/hhalpin/homepage/notes/vcardtable.html

[14]
http://www.nda.ie/cntmgmtnew.nsf/0/07AECF56B39CF6BB80257066005369
D8?OpenDocument

[15]
http://www.w3.org/TR/rdf
-
schema/

[16]
http://jena.sourceforge.net/

[17]
http://www.3ds.com/products/catia/portfolio/catia
-
v6/catia
-
v6r2011/#vid1