113
Touchpad as interaction input control for use of In-Vehicle Infotainment System Stefan Norberg 13th August 2009
Touchpad as interaction input control for use of
In-Vehicle Infotainment System
Stefan Norberg
13th August 2009
Touchpad as interaction input control for use of In-VehicleInfotainment System
Copyright © 2009 by Stefan NorbergPublished and distributed by:Department of Product and Production DevelopmentDivision Design & Human FactorsChalmers University of TechnologySE-412 96 Goteborg, Swedenwww.chalmers.se telephone: +46 (0)31-772 1000Printed in Sweden by Chalmers Reproservice Goteborg 2009
Abstract
Use of touch technology is increasingly used in handheld devices such as tele-phones and portable media players. Touch technology will also likely be a partof integrated in-vehicle solutions where more degrees of freedom are needed thantraditional interfaces provide.
However, there is considerable evidence that complex In-Vehicle Infotain-ment Systems (IVIS) with complex interactivity such as web browsing can dis-tract the driver. These systems will require new interaction methods betweenthe driver and the system. Such methods should allow tasks to be performed us-ing quick bursts of simple actions – without distracting the driver – yet powerfulenough to handle multitasking, panning, zooming and other possible interfaceprocedures.
We report on the development and evaluation of a multi-touch controlledin-vehicle infotainment system that allows for unrestricted Internet use, withfeatures that makes it adaptable to future applications.
In a simulator study and in a real driving test we assess its suitability forinteracting with a menu based on-board system. The system utilises multi-touch gestures to enable direct access to actions such as scrolling, zooming andpanning to provide an efficient and user-friendly interface.
In summary the results show that a rich multi-touch controlled interface canbe developed that users accept and like, regardless of previous personal pref-erence of touchpad usage. In addition the results indicate that a productionready touchpad controlled IVIS-system should be built with a bi-modal feed-back system, either visual-audio or visual-haptic to facilitate necessary drivercontrol with regard to road safety.
Keywords: HMI (Human Machine Interaction), Touchpad, IVIS (In VehicleInfotainment Systems), automotive, interface design, multi-touch, gestures, us-ability, multi-modal
Preface
During 2008-2009 this thesis was carried out by Stefan Norberg as the final partof the Masters program in Interaction Design at Chalmers University of Tech-nology in Goteborg, Sweden.
Volvo Cars Corporation AB, the OPTIVe (Optimized system Integration forsafe Interaction In Vehicles) project and Interaktionsbyran AB in Goteborg wereinitiators of the thesis.
ii
Thanks
I would like to send an appreciation to my supervisor and examiner Ulrike Raheat Chalmers University of Technology and supervisor Azra Moric at Volvo CarsCorporation.
I also want to thank Erik Hallgren and Olof Svensson for acting as opponentsduring the thesis presentation.
From Interaktionsbyran AB I would like to thank Emma Rozada, Joel Sand-strom, Anders Arnqvist, Daniel Werjefeldt, Henrik Andreasson for their helpand time.
At Volvo Cars Corporation I would like to thank, Mikael Gordh, Patrik Palo,Louise Rutgersson, Johannes Agardh, Ingrid Petterson, Robert Brostrom, Davidde Val, Peder Fast, Annie Rydstom, Samuel Palm and Niklas Hansson for theirhelp and engagement in the thesis.
Further I would like to thank all the test participants for their time and theireffort invested in the experiments conducted during the thesis.
This thesis was carried out in the OPTIVe (Optimized system Integration forsafe Interaction In Vehicles) project, financed by the Swedish IVSS (IntelligentVehicle Safety Systems) research foundation and the Swedish Road Administra-tion, which the author would like to thank.
Last but not least I want to thank my family and relatives for love andsupport.
iii
“Imagination is more important than knowledge. For knowledge is limited toall we now know and understand, while imagination embraces the entire world,and all there ever will be to know and understand.” - Albert Einstein
iv
Contents
1 Introduction 11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Thesis outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2.1 Chapter 1 - Introduction . . . . . . . . . . . . . . . . . . . 21.2.2 Chapter 2 - Methods . . . . . . . . . . . . . . . . . . . . . 21.2.3 Chapter 3 - Theory . . . . . . . . . . . . . . . . . . . . . . 21.2.4 Chapter 4 - Technology . . . . . . . . . . . . . . . . . . . 21.2.5 Chapter 5 - Realisation . . . . . . . . . . . . . . . . . . . 21.2.6 Chapter 6 - Result . . . . . . . . . . . . . . . . . . . . . . 31.2.7 Chapter 7 - Discussion and conclusions . . . . . . . . . . 3
1.3 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41.4 Interaction design . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.5 On studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71.6 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81.7 Goal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91.8 Research questions . . . . . . . . . . . . . . . . . . . . . . . . . . 91.9 Delimitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2 Methods 112.1 Literature study . . . . . . . . . . . . . . . . . . . . . . . . . . . 122.2 Store browsing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122.3 Youtube reviews . . . . . . . . . . . . . . . . . . . . . . . . . . . 122.4 Unstructured interviews . . . . . . . . . . . . . . . . . . . . . . . 122.5 Semi-structured interviews . . . . . . . . . . . . . . . . . . . . . . 132.6 Brainstorm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132.7 Checklists . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142.8 Ranking and weighing . . . . . . . . . . . . . . . . . . . . . . . . 142.9 Questionnaires . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142.10 Expert appraisal . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3 Theory 173.1 Design principles . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.1.1 Fitt’s Law . . . . . . . . . . . . . . . . . . . . . . . . . . . 173.1.2 Hick’s law . . . . . . . . . . . . . . . . . . . . . . . . . . . 183.1.3 Gestalt laws . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.1.3.1 Proximity . . . . . . . . . . . . . . . . . . . . . 193.1.3.2 Continuity . . . . . . . . . . . . . . . . . . . . . 193.1.3.3 Whole-part relationship . . . . . . . . . . . . . . 19
v
CONTENTS CONTENTS
3.1.3.4 Similarity . . . . . . . . . . . . . . . . . . . . . . 193.1.3.5 Closure . . . . . . . . . . . . . . . . . . . . . . . 20
3.1.4 80/20 design . . . . . . . . . . . . . . . . . . . . . . . . . 203.1.5 Iconic representation . . . . . . . . . . . . . . . . . . . . . 203.1.6 Usability . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.1.6.1 Effectiveness . . . . . . . . . . . . . . . . . . . . 213.1.6.2 Efficiency . . . . . . . . . . . . . . . . . . . . . . 213.1.6.3 Satisfaction . . . . . . . . . . . . . . . . . . . . . 213.1.6.4 Pleasure . . . . . . . . . . . . . . . . . . . . . . 21
3.2 Driving . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213.2.1 Driving and inattention . . . . . . . . . . . . . . . . . . . 22
3.3 Modalities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233.3.1 Visual . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.4 Display position . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243.4.1 Low display position . . . . . . . . . . . . . . . . . . . . 263.4.2 High display position . . . . . . . . . . . . . . . . . . . . 26
3.5 Controls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273.5.1 Mapping . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273.5.2 Indirect interfaces . . . . . . . . . . . . . . . . . . . . . . 273.5.3 Direct interfaces . . . . . . . . . . . . . . . . . . . . . . . 28
3.6 Input controls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303.6.1 Touchpad . . . . . . . . . . . . . . . . . . . . . . . . . . . 303.6.2 Touch screen . . . . . . . . . . . . . . . . . . . . . . . . . 313.6.3 Rotary switch . . . . . . . . . . . . . . . . . . . . . . . . . 323.6.4 Buttons . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333.6.5 Joystick . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.7 Menus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343.7.1 Menu design . . . . . . . . . . . . . . . . . . . . . . . . . 343.7.2 Menu interaction . . . . . . . . . . . . . . . . . . . . . . . 343.7.3 Alternative menus . . . . . . . . . . . . . . . . . . . . . . 35
3.8 Text input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353.8.1 1- and 2-dimensional keyboards . . . . . . . . . . . . . . . 353.8.2 Handwriting recognition . . . . . . . . . . . . . . . . . . . 35
3.9 Multi-functional control placement . . . . . . . . . . . . . . . . . 363.9.1 Centre armrest . . . . . . . . . . . . . . . . . . . . . . . . 363.9.2 Door-side . . . . . . . . . . . . . . . . . . . . . . . . . . . 363.9.3 Distributed . . . . . . . . . . . . . . . . . . . . . . . . . . 373.9.4 Steering wheel . . . . . . . . . . . . . . . . . . . . . . . . 37
3.10 Dominant hand . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4 Technology 394.1 Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 394.2 Capture program . . . . . . . . . . . . . . . . . . . . . . . . . . . 404.3 Test vehicle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 404.4 Driving simulator . . . . . . . . . . . . . . . . . . . . . . . . . . . 404.5 Touchpad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
vi
CONTENTS CONTENTS
5 Realisation 455.1 First iteration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.1.1 Knowledge gathering . . . . . . . . . . . . . . . . . . . . 465.1.1.1 Drivers . . . . . . . . . . . . . . . . . . . . . . . 465.1.1.2 Volvo Product Design . . . . . . . . . . . . . . . 475.1.1.3 Volvo Human-Machine-Interaction . . . . . . . . 475.1.1.4 Volvo Product Planning . . . . . . . . . . . . . . 47
5.1.2 Brainstorm . . . . . . . . . . . . . . . . . . . . . . . . . . 485.1.3 Checklist . . . . . . . . . . . . . . . . . . . . . . . . . . . 495.1.4 Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.1.4.1 Concept 1 . . . . . . . . . . . . . . . . . . . . . 495.1.4.2 Concept 2 . . . . . . . . . . . . . . . . . . . . . 505.1.4.3 Concept 3 . . . . . . . . . . . . . . . . . . . . . 505.1.4.4 Concept 4 . . . . . . . . . . . . . . . . . . . . . 53
5.1.5 Weighting . . . . . . . . . . . . . . . . . . . . . . . . . . . 545.1.6 Discussion and conclusion . . . . . . . . . . . . . . . . . . 54
5.2 Second iteration . . . . . . . . . . . . . . . . . . . . . . . . . . . 565.2.1 Store browsing . . . . . . . . . . . . . . . . . . . . . . . . 565.2.2 Simulator test . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.2.2.1 Prototype . . . . . . . . . . . . . . . . . . . . . . 575.2.2.2 Result . . . . . . . . . . . . . . . . . . . . . . . . 58
5.2.3 Graphic design . . . . . . . . . . . . . . . . . . . . . . . . 595.2.4 Driving test . . . . . . . . . . . . . . . . . . . . . . . . . . 595.2.5 Discussion and conclusion . . . . . . . . . . . . . . . . . . 61
5.3 Third iteration . . . . . . . . . . . . . . . . . . . . . . . . . . . . 615.3.1 Graphic design . . . . . . . . . . . . . . . . . . . . . . . . 615.3.2 The final concept . . . . . . . . . . . . . . . . . . . . . . . 615.3.3 Questionnaire . . . . . . . . . . . . . . . . . . . . . . . . . 625.3.4 Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . 625.3.5 Discussion and conclusion . . . . . . . . . . . . . . . . . . 66
6 Result 676.1 Interaction model . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
6.1.1 Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . 676.1.2 Gestures . . . . . . . . . . . . . . . . . . . . . . . . . . . . 706.1.3 Enter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 716.1.4 Home button . . . . . . . . . . . . . . . . . . . . . . . . . 716.1.5 Character input . . . . . . . . . . . . . . . . . . . . . . . 726.1.6 Simple questions . . . . . . . . . . . . . . . . . . . . . . . 72
6.2 Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 726.2.1 Home view . . . . . . . . . . . . . . . . . . . . . . . . . . 746.2.2 Contacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . 746.2.3 Navigation . . . . . . . . . . . . . . . . . . . . . . . . . . 746.2.4 Communication . . . . . . . . . . . . . . . . . . . . . . . . 776.2.5 Entertainment . . . . . . . . . . . . . . . . . . . . . . . . 776.2.6 Internet . . . . . . . . . . . . . . . . . . . . . . . . . . . . 776.2.7 Car settings . . . . . . . . . . . . . . . . . . . . . . . . . . 806.2.8 Favourites . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
6.3 Controls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
vii
CONTENTS CONTENTS
7 Discussion and conclusion 837.1 Further studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 857.2 Personal recommendations . . . . . . . . . . . . . . . . . . . . . . 86
A Interview questions 91
B Questionnaire 93
viii
List of Figures
1.1 Worldwide Mobile Phone/Touch Screen Mobile Phone MarketForecast [Displaybank, 2009] . . . . . . . . . . . . . . . . . . . . 4
1.2 Functions available in Volvo cars 2008 [Bergmark, Gustafsson, 2008] 51.3 Screen-shot of Cover Flow in Apple iTunes 8.1 (OS X version) . . 5
2.1 Design process - overview of phases [Jones, 1992] . . . . . . . . . 11
3.1 1- and 2-Dimensional Fitt’s Law test . . . . . . . . . . . . . . . . 183.2 Plot of Fitt’s Law where a=0.23, b=0.166, (d,s) = [0-5] inches
(plotted in Matlab by the author) . . . . . . . . . . . . . . . . . . 193.3 Gestalt laws of perception [Benyon et. al., 2005, p. 114] . . . . . 203.4 Examples of iconic representation (work of the author) . . . . . . 203.5 Percentage of events for attention by severity level [Dingus et al., 2006,
Figure RO.13, p. xli ] . . . . . . . . . . . . . . . . . . . . . . . . 223.6 Frequency of Occurrences of Crashes and Near Crashed[Dingus et al., 2006,
Figure RO.14, p. xlii] . . . . . . . . . . . . . . . . . . . . . . . . 233.7 Glance data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243.8 Different fields of vision [Bergmark, Gustafsson, 2008] . . . . . . 253.9 Display position in degrees [Bergmark, Gustafsson, 2008] . . . . 263.10 Example of mapping . . . . . . . . . . . . . . . . . . . . . . . . . 283.11 Examples of indirect interfaces . . . . . . . . . . . . . . . . . . . 293.12 Examples of direct interfaces . . . . . . . . . . . . . . . . . . . . 293.13 Example of touchpad with two buttons and static scroll bars . . 303.14 Relative and absolute selection style on touchpad . . . . . . . . 313.15 Example of touch screen displaying a user interface . . . . . . . 323.16 Example of rotary switch (BMW iDrive) . . . . . . . . . . . . . . 323.17 Example of buttons . . . . . . . . . . . . . . . . . . . . . . . . . . 333.18 Example of joystick (Lexus Remote Touch) . . . . . . . . . . . . 343.19 Example of pie menu [Kurtenbach, 1993] . . . . . . . . . . . . . . 353.20 Example of 1- and 2-dimensional keyboards . . . . . . . . . . . . 36
4.1 Overview of prototype setup . . . . . . . . . . . . . . . . . . . . . 394.2 Detailed overview of prototype setup . . . . . . . . . . . . . . . . 404.3 Volvo XC90 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414.4 Lindholmen Open Arena driving simulator . . . . . . . . . . . . . 424.5 Driving simulator interior . . . . . . . . . . . . . . . . . . . . . . 424.6 Apple iPod Touch . . . . . . . . . . . . . . . . . . . . . . . . . . 43
ix
LIST OF FIGURES LIST OF FIGURES
5.1 Concept 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 505.2 Concept 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515.3 Concept 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 525.4 Concept 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 535.5 Gestures and increasing dimensions of freedom . . . . . . . . . . 555.6 Selection model concept . . . . . . . . . . . . . . . . . . . . . . . 565.7 Mood board for the graphic design . . . . . . . . . . . . . . . . . 605.8 Concept average . . . . . . . . . . . . . . . . . . . . . . . . . . . 635.9 Concept evaluation: Coefficient of variation . . . . . . . . . . . . 645.10 Touchpad average . . . . . . . . . . . . . . . . . . . . . . . . . . 645.11 Touchpad average: Coefficient of variation . . . . . . . . . . . . . 655.12 Prototype appreciation average as function of touchpad average . 65
6.1 Touchpad with click . . . . . . . . . . . . . . . . . . . . . . . . . 676.2 Proposed and tested touchpad mounting position . . . . . . . . . 686.3 Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 686.4 Selection in fine mode . . . . . . . . . . . . . . . . . . . . . . . . 696.5 Selection in coarse mode . . . . . . . . . . . . . . . . . . . . . . . 696.6 Selection model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 706.7 Gestures in the final concept . . . . . . . . . . . . . . . . . . . . 716.8 Touchpad and home button . . . . . . . . . . . . . . . . . . . . . 726.9 Character input dialogue . . . . . . . . . . . . . . . . . . . . . . . 736.10 Simple questions . . . . . . . . . . . . . . . . . . . . . . . . . . . 736.11 Home view . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 756.12 Contacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 756.13 Contacts as a list . . . . . . . . . . . . . . . . . . . . . . . . . . . 766.14 Multiple types of information and actions available for a contact 766.15 Navigation interface displaying a sample map . . . . . . . . . . . 776.16 Communication view . . . . . . . . . . . . . . . . . . . . . . . . . 786.17 Entertainment view displaying a number of record covers . . . . 786.18 Internet view with dn.se loaded and the first article selected . . . 796.19 Internet view with page transition in progress . . . . . . . . . . . 796.20 Internet view with article zoomed in . . . . . . . . . . . . . . . . 80
x
List of Tables
3.1 Low display summary . . . . . . . . . . . . . . . . . . . . . . . . 263.2 High display summary . . . . . . . . . . . . . . . . . . . . . . . . 273.3 Touchpad summary . . . . . . . . . . . . . . . . . . . . . . . . . 313.4 Touch screen summary . . . . . . . . . . . . . . . . . . . . . . . . 313.5 Rotary switch summary . . . . . . . . . . . . . . . . . . . . . . . 323.6 Button summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 333.7 Joystick summary . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.1 Comparison of concepts . . . . . . . . . . . . . . . . . . . . . . . 54
xi
LIST OF TABLES LIST OF TABLES
Word list and list of abbreviations
Affordance Action possibilities which are readily perceivable by an actor
CVF Central Vision Field
FVF Functional Vision Field
Gestalt An organised whole that is perceived more than the sum of its parts
HCI Human Computer Interaction
HMI Human Machine Interaction
IVIS In Vehicle Infotainment System
Infotainment Broadcast material that is intended for information and enter-tainment
PVF Peripheral Vision Field
PPL Product Planning, strategic planning
Stakeholder The various parts with an interest in a product
xii
Chapter 1
Introduction
1.1 Introduction
Drumming on the wheel, John is quite annoyed on the authorities for not finish-ing the bridge that was supposed to rid the big traffic jam now building up infront of him. This day he doesn’t have time to get stuck. The ferry to Gdansk,Poland, leaves in two hours and he has to pick Eva up soon if not to fall behind.An inch too tight schedule, he admits.
John finally passes the jam and drives onto the freeway and speeds up, butafter a while he realises that he has taken the wrong way; he’s heading towardsEva’s former workplace.
- OK, what to do, he ponders.He looks at the display in the car and realises that Eva used the navigation
system in the car to find the address to her new workplace a few days ago - Away point called “New office” is saved in the favourites tab on the right.
He moves his finger to the right on the touchpad and clicks it, a route toEva’s office is instantly presented to him on the display and a small note downin the corner informs him, “Showing best alternate route, fastest route blockedby traffic jam”
While exiting the freeway on the next ramp he calls Eva, again using thefavourites tab. After a few signals Eva answers.
- This is Eva.- I’ll be outside your office in, hmmm - he looks at the display and reads “9
minutes to New Office” - 10 minutes, he says.- I’ll get my things and come down to meet you outside the reception then.- OK, bye.John drives on and looks out trough the window, while the light sand-
coloured concrete of the elevated highway passes beneath him in a steady pace.When he reaches the top of the arch that the slowly winding highway creates,he sees a massive traffic jam to the North. That must be the traffic jam thesystem was telling me about, he thinks to himself while he follows the drivingdirections on the display, guiding him trough a drowsy suburb with well-kepthomes. I would never have driven here if the system had not told me to, hethinks.
After 10 minutes of driving trough unexplored suburban territory he arrives
1
1.2. THESIS OUTLINE CHAPTER 1. INTRODUCTION
at Eva’s office, and she comes out from the left side, of the large shiny glassfacade that frames building.
Now, both in the car and on schedule, John asks Eva to pump up some music,maybe ”Classic Mix Radio, a favourite web-station on my behalf”, he suggests.Eva, while not a lot familiar with the system in front of her still recognises theradio-symbol on screen, and as it pops open the station John asked for is in thefirst list. She scrolls a few steps and music is on.
- ”I’ve found us a nice stay with view over the old city in Gdansk, there’sdinner at eight”, John says.
- ”Great, that leaves time for some city strolling, does this system haveGoogle in it?”
-”Sure.”Eva brings up the menu where she had found the radio, Internet search is
there, soon she is visually prompted for character input. ”GDANSK SIGHT-SEEING”, she enters and up comes the regular hit list. ”How do I zoom in?””Oh, just like this”, says John and moves his hand. ”But please, before, couldyou just put on an album of Madonna or something, I have her latest, this stuffis boooooring me.” Eva is a little amazed of how fast she’d nail it down, butMadonna is on and she is again looking at a tourist-page of Croatia.
1.2 Thesis outline
1.2.1 Chapter 1 - Introduction
An introduction and overview of the project with general information aboutthe motivation and background to conduct the research are presented here. Itcontains the initial research questions and delimitations that were set up in thebeginning of the project.
1.2.2 Chapter 2 - Methods
The methods used to carry out the research and the work to create the proto-types at different levels throughout the project are described in this chapter. Themethods that were used to develop and design the questionnaires, the strategiesto analyse the result and collect data from the participants are presented here.
1.2.3 Chapter 3 - Theory
This chapter contains the theoretical framework that was used as support tothe work of creating the concept. Theory from this chapter was used as guideand support to make design decisions.
1.2.4 Chapter 4 - Technology
This chapter contains an overview of the technology that was used in the project
1.2.5 Chapter 5 - Realisation
This chapter describes the systematic process of putting the theoretical andconceptual parts together to develop the prototype, as well as the analysis of
2
CHAPTER 1. INTRODUCTION 1.2. THESIS OUTLINE
the test results. It describes how the work was carried out, what hardshipsdid occur, what design decision were made and why they were made. Whichquestions were considered and how they were handled.
1.2.6 Chapter 6 - Result
The result of all work is presented here.
1.2.7 Chapter 7 - Discussion and conclusions
The lessons learned and what the results imply. Further fields of study arepresented here.
3
1.3. BACKGROUND CHAPTER 1. INTRODUCTION
Figure 1.1: Worldwide Mobile Phone/Touch Screen Mobile Phone Market Fore-cast [Displaybank, 2009]
1.3 Background
Use of touch technology has during the last 20 years gained user acceptance ina variety of consumer applications such as touchpads in laptops and portablemedia players [Hsu, 2009]. According to ABIresearch the mobile touchscreenmarket will be worth $5 Billion dollars in 2009 [ABIresearch, 2008]. If thisis a true forecast or not remains to be seen, but other sources such as themarket research firm Displaybank, expects an increase in touch technology (seeFigure 1.1) used in mobile phones, from 4% (52 million units) 2008 to 10% (114million units) 2009 [Displaybank, 2009].
Touchpads are viable to let users access more degrees of freedom, to con-trol applications such as Internet navigation, where use of traditional interfacesare inefficient, while a rapidly increasing number of new information and en-tertainment systems have been introduced to the automotive industry over thepast few years. The number of functions from 1998 to 2008 have increasedwith 218% in Volvo cars, most divided between infotainment (IVIS) and act-ive safety (ADAS) (see Figure 1.2 on the next page), and the increased flowof information has a cognitive affect on the drivers attention, mental workloadetc.[Bergmark, Gustafsson, 2008]. The use of Internet has also got increasinglymore popular and there is an expressed customer need for Internet availabilityin commercial vehicles.
Primarily this will require an interface navigation principle, simple enough toperform tasks in quick simple bursts of actions – without distracting the driver– yet powerful enough to handle multitasking, panning, zooming and otherpotential interface procedures. Road safety is still supreme in car industry andsystems must be designed not to distract drivers.
Developing a solution that does not benefit the drivers is pointless, since thedrivers are the buyers and the source of revenue for the company. Therefore,the scope for this thesis will be to search for a solution that the user findspleasurable to use and efficient with regard to safety. Focus will be placed onhuman- and user-centred qualities to find a suitable solution, for example, the80/20 design rule (see 3.1.4 on page 20) will be used; which predicts that 80percent of the time in any large system is spent using 20 percent of the features,and the system should be designed for efficient use of these functions.
4
CHAPTER 1. INTRODUCTION 1.4. INTERACTION DESIGN
Driver
Radio, CD,
DVDMobile
phone
PDAInternet,
Road and tra!c
information
Speed alert
Vehicle
diagnostics
Driver impairment
monitoring
Vision
enhancement
Safe
following
IVIS
ADAS
Collision
warning
Lateral control
support
Active pedestrian
protection
Figure 1.2: Functions available in Volvo cars 2008 [Bergmark, Gustafsson, 2008]
Figure 1.3: Screen-shot of Cover Flow in Apple iTunes 8.1 (OS X version)
1.4 Interaction design
An example that illustrates how interaction design affects a person is the actof flickering trough a “cover flow” (see Figure 1.3). At first intuition and gutfeeling gives the immediate though, that it looks a bit like a physical collectionof records. The action of flicking trough the covers can be done with a sweepingfinger movement. When you get to the end, it acts elastic, like stretching arubber band, to communicate that you have reached the end of the list. After awhile after you may think that its just a software feature, but it is a clever wayof showing you that there are no more items in the list. The similarity with arubber band is striking but yet subtle, elasticity, an interaction synonym for thereal physical property of something that is stuck or ending.
I own an old Mercedes 1986 560 SEL, it is a pleasure to drive, its smooth,fast, strong, really well designed and it leaves an impression of quality. It is al-most perfect except for one thing; it lacks a cup holder, probably because therewere very little drive-in-coffeehouses in 1986 or maybe because the designersthought that you should stop for refreshments because of safety reasons. Thecar was built with focus on safety, style, simplicity, features and function ina wonderful combination. It has 8-way adjustable ventilated power seats withdriver memory, 4-way adjustable power rear seats, automatic self-levelling sus-pension and courtesy lights underneath the doors light up the ground when youstep in or out of the car.
The interaction with the functions in the car is very well thought trough.
5
1.4. INTERACTION DESIGN CHAPTER 1. INTRODUCTION
For example if the windshields or the mirrors are foggy or icy the driver canpress the “heat windows and mirrors”-button which automatically turns up theheat, turns on the thermostat control in the mirrors and redirects the airflowto the windows. When the ice or fog is gone, the driver can press the “normal”-climate mode button and everything goes back to the previous setting, with atemperature setting between 18-27 Cº, an optimal range that fits most people.There are many features (although not as many as in a more recent car), butthey are arranged in a way that gives the driver control over the most important.The rest is automatic or cannot be changed, and there is no reason why anyonewould ever want to change it, they should busy enjoying the driving instead.
How can parallels be drawn between this car and a newer Mercedes S600?The same company is building it, its marketed as one of the most well designed,well engineered and most luxurious cars on can get. It boasts a massive numberof functions, of which many are seemingly unnecessary, for example, the timethe numberplate light should stay on, after the car has been turned off, canbe adjusted [BBC, 2008]. I wonder if any customers has been concerned withthat or mentioned it as a problem? This is exaggerating a bit, but the point is,reflection.
Actually, this particular feature was probably a product of the organisa-tional structure of a large technology driven car company. The engineers thatconstructed the car need to know this information of course, but why did it endup as a feature that the driver can adjust?
Perhaps a credible scenario would have been a meeting, where two differentteams could not agree what to do, what features to select and why they should beselected, and as a result this feature was exposed to the driver as a precaution notto fall behind competitors. A better alternative would have been to approachthe problem from the drivers perspective and understand their experience ofdriving and buying the car. Without any strategy or process to work with theusers this could easily be overlooked and decisions can be hard to take withoutguidelines.
This example might not seem that bad, but every feature needs to be imple-mented, tested, verified and the user interface has to be designed to allow thedriver to access this feature, while the customer gets a system with a featurethat they (probably) never will benefit.
Engelbrektsson and Soderman discusses the discrepancy between the notionof having a dialogue with the users during the development process, and actu-ally working active with user-centredness and user requirements in their article“The use and perception of methods and product representations in productdevelopment: A survey of Swedish industry” [Engelbrektsson et al., 2004].
They gave about 1000 Swedish companies statements like “It is importantto have continuous communication with the customers during the product de-velopment process” and “The early analysis of customer requirements is oftenneglected in order to save time” to investigate how companies handled user re-quirements. Their results indicate for example that early analysis of customer re-quirements is often neglected in order to save time, even though the respondentsassert the importance of customer communication. In practise product develop-ment processes tend to be focused on time-related efficiency rather that customeroriented activities for improved product quality [Engelbrektsson et al., 2004].
Engelbrektsson and Soderman found in particular that companies using Fo-cus Groups and Observation Studies perceived that they discovered fewer re-
6
CHAPTER 1. INTRODUCTION 1.5. ON STUDIES
quirements late in the development cycle than others [Engelbrektsson et al., 2004].Another finding was that companies using Rapid Prototyping discovered fewerproduct requirements in the late stage of the product development cycle anda possible explanation is that this method can speed up development and al-low several cycles, which makes it possible to test and evaluate products morefrequently, while still in development [Engelbrektsson et al., 2004].
I think user-centredness and knowledge of how to handle user-requirementsis an important issue to address for interaction design, since it often is focusedon software within industry. Software problems might be hard see before theproduct is almost completely finished, since software specification often are bigand cumbersome to read. Instead the problems end up in an almost finishedproduct, which makes them harder to fix than if the problems were addressedearlier in the cycle. By gathering requirements from the users in the beginning ofthe development cycle, this could be avoided to some extent as Engelbrektssonand Soderman found in their study.
1.5 On studies
There might be a need to consider how studies are designed to make sure thatvaluable results are obtained. If one wants to know which interface is the mostsuitable for a task, both the interaction method and the interface design needsto be considered, as well as some information of the context of the problem, thatone would like to solve. For example comparing a method to select an alphabeticelement (e.g. a name) in a list using a touchpad and a rotary controller couldbe designed in several different ways. A trivial approach would be to use linearselection method on the touchpad and ditto for the rotary controller. That testwould not give much valuable information without any knowledge about thecontext of use, even if results would be obtained.
For example, if the list is very long, the first letter of an element can beinterpreted by the touchpad, drawing it with a finger, thus providing a shortcutin the list, giving the touchpad an advantage. Respectively, the rotary controllercan be equipped with two-speed scrolling, one very fast to traverse long lists andone slow to select the correct element, levelling the distance in efficiency, etc.
Many studies seem to ignore this reasoning and compare interfaces and in-teraction methods with unjust advantages, that favours a particular result orseverely limits other concepts potential.
For example one study compares the speed of alphanumeric character in-put with a thumb-controlled touchpad, mounted on the steering wheel in a carsimulator. A 1-D linear keyboard, 2-D keyboard and sign-recognition was com-pared to each other, and the conclusion was that the 2-D keyboard was theslowest[Gonzalez et al., 2007, p. 102]. “A possible explanation for this is thefact that our steering wheel prototype was not optimised for thumb movementalternating along both axes” [Gonzalez et al., 2007].
The experiment was set up to compare three different interaction techniquesbut unfortunately the setup crippled the user, since they could not use theirthumb as intended. Results like that needs to be considered when designingand testing interfaces and input methods, to make sure that the data obtainedfrom the experiment will be useful.
Other studies do not show how their interfaces were designed, which is in-
7
1.6. PURPOSE CHAPTER 1. INTRODUCTION
formation at least as important as knowing what interaction method was used,from a point of interaction design, but it has natural explanations. All studiesinteresting from a point of interaction design are not necessarily conducted bypeople that have a designer background. Many evaluations are of a technicalcharacter and therefore not concerned with interface design, usability or use-fulness. Research papers are sometimes required to fit a specific formats andshowing pictures of the interfaces can be space consuming, which might be alimitation. Describing the behaviour of an application in text is not easy and itcan be very hard to follow.
A study evaluating multi-functional systems available in a number of vehicles,found that many usability problems have a lot to do with the graphical solutionsand less to do with the actual manual interaction principle, which emphasisesthis issue[Rydstrom et al. 2005].
1.6 Purpose
The main purpose of this thesis is to investigate the potential of a touchpadas the main interaction principle for IVIS. A conceptual design solution willbe developed and evaluated to establish how well the concept can handle thechallenges that comes with the next generation of infotainment systems. Asubjective evaluation to investigate the user acceptance towards the technologywill also be performed.
Other factors of importance are:
• Research indicates that there is room for improvement with new interac-tion models, because of the increasing number of features that are pushedinto the cars
• A well executed interaction concept can become an important key propertythat defines the car and the company [Gordh, 2008]
Further specified, the aim is to develop, prototype and evaluate the followingscenarios in one sequence, using touchpad technology:
• Make Call > Look at Map > End Call
• Start Radio > Quick Web Search > Find Artist Album.
Including actions:
• Change between several sources (Radio, Media, Phone, Navigation etc)
• Start a destination-guide by putting a needle on the map
• Activate preferred Radio station (by saved presets, manual tuner or re-cently played).
• Find a specific music artist and shuffle songs.
• Answer the incoming phone and end the phone call
• Google search for traffic and tourist information on a web page
8
CHAPTER 1. INTRODUCTION 1.7. GOAL
1.7 Goal
The goal is to create a functional prototype that can be evaluated to measureuser acceptance of touchpad usage for IVIS.
Usability challenges/prerequisites The usability challenges to be over-come for the safe use of IVIS are:
• The concept must be usable to both novice and expert uses
• The concept must be usable by drivers of different ages and differentlengths
• The concept must be usable for both left- and right hand driving
• The concepts shall not interfere with safe execution of the driving task. Itshall not distract, disturb or overload drivers
• The placement of device/control is important. The driver must not acci-dentally activate control and it should not lead to unnecessary distractionfrom the forward view when the driver has actuate a control
1.8 Research questions
The initial research questions are:
• How could a new touchpad interface be designed to decrease distractionand to gain user acceptance?
• What is the best way to interact with a touchpad?
• Which more complex actions are needed beyond clicking and scrolling?
• Which tasks raise/ decrease complexity when performed with touchpad?
• How can multitasking be treated?
1.9 Delimitations
• Interaction method (control interface) is in focus and not the features ofthe system (e.g. should the system have CD/Satellite Radio or CD/Webradioor both, etc.)
• Focus is primary on driver interaction while driving and not interactionfrom a potential passenger
• The study will not elicit new functions for IVIS
• Voice controlled interfaces shall be excluded
• Levels of distraction will not be examined
• Techniques for distinguishing input between the driver and passenger willnot be subject for this thesis
9
1.9. DELIMITATIONS CHAPTER 1. INTRODUCTION
• Master thesis shall not examine possibilities to reformatted and limit webpages, even though simplified browser might be suitable for an In-vehicleHMI while driving. The assumption can be made that everything on webis available or accessible while driving. The challenge is to make sure thatinteraction still is preformed in the most efficient and safe way
• It is not within the scope of the thesis to evaluate driver or task perform-ance
10
Chapter 2
Methods
In this part different methods to elicit and collect data that will construct thefoundation and the theoretical framework for this report will be discussed.
The design process that was used can be found in book called Design Meth-ods by John Chris Jones[Jones, 1992, p.66]. The process can roughly be de-scribed as three different states, divergence, transformation and convergence asseen in Figure 2.1. Each of these states has their own specific methods and theirown purpose to distill and aid the designer in the systematic and artistic work.
A brief overview of which different methods were used in the different phases:
1. Divergence - Determination of problem
• Literature study
• Unstructured interviews
• Store browsing
• Youtube reviews
• Brainstorm
2. Transformation - Systematic handling of problem
• Checklist
• Rapid prototyping
• Weighting and ranking
• Expert evaluation
Divergence
Transformation
Transformation
Transformation
Convergence
Figure 2.1: Design process - overview of phases [Jones, 1992]
11
2.1. LITERATURE STUDY CHAPTER 2. METHODS
• User workshop and testing
3. Convergence - Evaluation and verification
• Questionnaire• User testing• Semi-structured interviews
2.1 Literature study
To gain understanding of the problem, some initial information will have tobe acquired and available literature and papers gives an overview and state-of-the-art. The areas to research includes human cognition, car safety, designguidelines, technical documents for implementation, human response times andproduct reviews of devices. This information area is larger than just touch-pads, but it should contain information, scattered and disorganised, that can begathered to elicit the requirements and theory necessary to create the concept[Jones, 1992, p. 202].
2.2 Store browsing
Gathering information about existing products can be helpful when introducedto a problem, for example, to determine the state-of-the-art, or research whatperformance can be expected from a particular range of products. There isusually a creative phase in the beginning of a project when many ideas aregenerated, which are built upon the designers experience or influences. In thisstage there might be uncertainties about what potential the ideas have and howthey can be formalised, because the designer has insufficient knowledge at thattime.
Knowledge of what different products are capable of gives the designer apersonal experience to get an idea about a particular interface or property.Initial thoughts about a design can be analysed and take shape when a product isexperienced in a real implementation, which can help justifying a design decision.
2.3 Youtube reviews
This might be considered an observation based method that is built on watch-ing Youtube movies, to learn how people perceive different products and sys-tems. Watching users try out and explain their impression of a product canbe beneficial, since they comment issues or features of a particular design. Themethod can be useful to identify so called performance requirements, which arerequirements that the users perceive as a problem or possible improvement ofan existing product [Karlsson, 2007].
2.4 Unstructured interviews
This type of interview is appropriate when the investigator has little idea ofwhat the issues of concern are. It gives the interviewer an overview and a basic
12
CHAPTER 2. METHODS 2.5. SEMI-STRUCTURED INTERVIEWS
understanding in the different problem areas, from the people experiencing theproduct.[Jordan, 2000, p. 159].
The interview is conducted with a series of open-ended questions, to get theconversation started, and that gives the respondents the possibility to steer theconversation towards areas they consider important.
2.5 Semi-structured interviews
This method is used when the investigator has a clearer idea of the relevant issuesto evaluate. A bit more constrained than unstructured interviews, this methodensures that specific questions are answered by the respondents so that a centralset of issues can be covered. The same questions can be prompted to all respond-ents, which gives a chance for more systematic analysis of their answers, whilerespondents can raise issues they find of particular importance.[Jordan, 2000, p.159]
2.6 Brainstorm
Brainstorm is a well known method that is practised in companies with varyingresult all over the world. It can properly be used generate and structure ideas inthe beginning or continuously throughout a project. Focus for this method is tospark creativity and use the synergy between participants to increase creativity.The overall structure can briefly be described as three parts; collect a numberof people; generate ideas; systematise the result [Lovgren et al., 2004, p. 93].
The participants should preferably have a relation to the project or the ques-tions posed, but mixing persons with different background can prove beneficialfor the result and increase the group dynamics. During the idea-generationphase, focus should be to come up with solutions to a relatively open-endedquestion, posed by a master-of-ceremony that guides the group (approximately3-7 people) through the session. The session is loosely bound to follow approvedrules (see below) [Kelley, 2006, p. 143].
• Go for quantity
• Encourage wild ideas
• Be visual
• Defer Judgement
• One Conversation at a time
Then at the end of the session the ideas will be grouped and classified to see thestructure and how different ideas relate. However, accordingly to Lovgren andStolterman there are no proofs, that the brainstorming method will generateresults of higher quality, than if every member spent the same time working onthe problem alone [Lovgren et al., 2004, p. 95]. Nevertheless, it will be usedincrease group dynamics and exchange ideas, which is plausible for example, inthe idea-generating stage of the project.
13
2.7. CHECKLISTS CHAPTER 2. METHODS
2.7 Checklists
Checklists are used as a “reality check” to make sure that generated concepts areusable and meet certain criteria [Jones, 1992, p. 362]. The criteria can eitherbe established in the beginning of a project (see 1.7 on page 9) or in later stagesfrom requirements collected from different stakeholders (see 5.1.3 on page 49).
2.8 Ranking and weighing
Choosing between different design concepts requires some structure and consid-eration with focus on the objectives that the designs are to satisfy. Ranking andweighing is used to compare a set of alternative designs using a common scaleof measurement.
1. Ranking
(a) Record on a matrix the preferred objects of each pair(b) Rank objectives in the order of their preference score
2. Weighing
(a) Assign an index number to each objective to indicate its importancerelative to the others.
(b) Measure or estimate the degree to which each alternative design sat-isfies each of the ranked or weighted values
(c) Convert these measures or estimates to percentages in the case ofranked objectives, and to value of the index numbers, in the case ofweighted objectives.
(d) Select the alternative design either having the preferred pattern ofpercentages or the highest total of weighing index numbers
Ranking and weighing can be done in different elaborate ways and combinations,but in this report a scale ranging from 1 to 3 is used, with 1 representing the leastweight and 3 representing the most. In combination with logic reasoning andchecklists (see 2.7) the result was used to determine what concepts to furtherdevelop after the first iteration (see 5.1 on page 45). Weighing of objectivescan arguably distort the pattern of the problem, because it is influenced by theexpectations one has on the objectives and how they are going to be achieved.By using a checklist one can be on the lookout for this weakness with thismethod and correct obviously erroneous results and instead reassess weights asneeded[Jones, 1992, p. 377].
2.9 Questionnaires
A questionnaire can be used to collect information from the members of a largepopulation when evaluating a certain property. Questionnaires are useful andfast when collecting data from a large number of test subjects, but a drawbackis their inflexibility and the sensitivity of how the questions are formulated. It isimportant to make pilot tests on questionnaires to make sure that the questionsare properly understood to avoid faulty data.
14
CHAPTER 2. METHODS 2.10. EXPERT APPRAISAL
2.10 Expert appraisal
An expert in this context is a person that has knowledge to make an informedjudgement on issues concerning the product or design under investigation. Pro-fessional training gives the expert special knowledge in areas that apply to theproduct or some other field that relates to the product e.g. demographics, er-gonomics, etc. Assessments from experts of different areas can be compared toget a cross-validation of the issue. No users are needed for this method and thespecialists knowledge should directly lead to a number of solutions. As no usersare involved there is no real evidence available; Kerr and Jordan report froman evaluation of a prototype phone system, that only five of twelve predictionsfrom experts were supported in a subsequent empirical evaluation. In contrasteleven out of twelve predictions gathered from potential users were supported[Jordan, 2000, p. 172]. However, expert generated information can be a bitdifferent that the one from users, but primarily the information is more detailed[Karlsson, 2007].
15
2.10. EXPERT APPRAISAL CHAPTER 2. METHODS
16
Chapter 3
Theory
3.1 Design principles
The principles and theory presented here was used when designing the interfaceand the interactions to control it. They worked as guidelines and was used assupport to aid and motivate design decisions.
3.1.1 Fitt’s Law
The time required to move to a target is a function of the target size and thedistance to the target. Fitt’s law is only valid for rapid pointing movements anddoes not apply for writing or continuous movements.
MT = a + b log2(d
s+ 1)
Where the equation consists of these values for humans utilising a pointingdevice such as a mouse or touchpad:
• MT = movement time to a target,
• a = 0.230 [s]
• b = 0.166 [s]
• d = distance between pointing device and target [inch]
• s = size of the target [inch].
Pointing movements typically consists of three parts; one quick movement to-wards the target (ballistic movement), followed by fine-adjustment movements(homing movements) to a resting position (acquiring) over the target. Homingmovements are generally responsible for most of the movement time and causeof errors [Lidwell et al., 2003, p. 82].
In interface design that utilises pointing devices this implies that commonlyused controls should be close and large and controls not commonly used shouldbe smaller and more distant. For example using the screen border to constrainmovements would imply that a button placed there has infinite size i.e. it isvery fast and easy to push [Lidwell et al., 2003, p. 82].
17
3.1. DESIGN PRINCIPLES CHAPTER 3. THEORY
Figure 3.1: 1- and 2-Dimensional Fitt’s Law test
A Fitt’s Law test (see Figure 3.1) requires the user to fast click one of thecoloured fields; after the user clicks, the colours of the fields are switched andthe process is repeated. Varying the size and distance between the fields makesit possible to calculate the constants a and b which differs depending on inputmethod e.g. touchpad, touch screen, mouse or stylus. See Figure 3.2 on thefacing page for a sample plot that shows a function surface of Fitt’s law.
3.1.2 Hick’s law
The time it takes to make a decision increases as the number of alternativesincreases (a. k. a Hick-Hyman Law). It can be described by the equation:
RT = a + b log2(n)
Where:
• RT = Response time [s]
• a = total time that’s not involved in decision making [s]
• b = 0.155 empirical determined for humans [s].
• n = the number of alternatives to choose from
Hick’s law is most applicable to simple decision making tasks in which thereis a unique response to each action, e.g. if A happens press button 1, and lessapplicable in cases where the task in more complex, like choosing in a menu withcomplex sub-menus. Keeping Hick’s law in mind when designing timecriticaltasks and minimise options helps preventing error and reduce response times.[Lidwell et al., 2003, p. 103]
18
CHAPTER 3. THEORY 3.1. DESIGN PRINCIPLES
Figure 3.2: Plot of Fitt’s Law where a=0.23, b=0.166, (d,s) = [0-5] inches(plotted in Matlab by the author)
3.1.3 Gestalt laws
In the beginning of the twentieth century a group of psychologists workedtogether and identified ’laws’ of perception that they regarded being innate.These laws, despite their age, are very useful for a number of interface designfeatures, those used in the concept are listed in Figure 3.3 on the next page[Benyon et. al., 2005, p. 114].
3.1.3.1 Proximity
Objects close in proximity in time or space tend to be perceived together, whichis useful to create a grouping or cluster with functions that logically relates toeach other.
3.1.3.2 Continuity
Smooth continuous patterns are perceived more easily than disjoint interruptedones. Figure shows continuity constructed from five semicircles. The figure tendto be perceived as a continuous curve instead of five semicircles.
3.1.3.3 Whole-part relationship
Figure 3.3 on the following page illustrates this. It appears to be one image butit is composed of many parts.
3.1.3.4 Similarity
Similar figures tend to be perceived as grouped together.
19
3.1. DESIGN PRINCIPLES CHAPTER 3. THEORY
Figure 3.3: Gestalt laws of perception [Benyon et. al., 2005, p. 114]
Figure 3.4: Examples of iconic representation (work of the author)
3.1.3.5 Closure
Closed figures are perceived more easily than incomplete or open figures.
3.1.4 80/20 design
A high percentage of effect in any large system are caused by a low percentageof variables (Pareto’s principle). The 80/20 rule is useful for focusing resourcesthat in return realises greater efficiencies in design. For example if 20 percent ofa products features are used 80 percent of the time, testing and design shouldfocus on these. The remaining features should be reevaluated to verify theirvalue in the design [Lidwell et al., 2003, p. 13].
3.1.5 Iconic representation
Using iconic pictorial images is an effective way to make actions, objects and con-cepts in a display easier to remember, recognise and learn (see Figure 3.4). Iconicrepresentation reduces performance load and makes more efficient use of displaysurface and it makes controls more understandable cross-culture [Lidwell et al., 2003,p. 111].
20
CHAPTER 3. THEORY 3.2. DRIVING
3.1.6 Usability
Usability is the concept of creating usable functional products, clearly a productwill be useless if it does not contain the functions necessary to perform the tasksfor which it is intended. The International Standards Organisation (ISO) definesusability as:
”the effectiveness, efficiency and satisfaction with which specifiedusers can achieve specified goals in a particular environments.”
(ISO DIS 9241-11)
During the years usability has evolved from what marketing people call ’satisfier’to ’dissatisfier’, which means that people nowdays expect the product to providethe right functionality and they become dissatisfied otherwise [Jordan, 2000].
3.1.6.1 Effectiveness
According to the ISO standard this is “the extent to which a goal, or task, isachieved” [Jordan, 2000].
3.1.6.2 Efficiency
ISO defines efficiency as“the amount of effort required to accomplish a goal”[Jordan, 2000].
3.1.6.3 Satisfaction
The level of comfort that the user feels when using a product [Jordan, 2000].
3.1.6.4 Pleasure
Usability is vital but it’s not the not the whole story. The development of usab-ility has pushed the limit further in the field, to a point where also the pleasureassociated with the product is included in the design. A holistic approach toaddress people and products gives an understanding of what role, beyond usab-ility, the product plays in peoples life [Jordan, 2000]. Taking this into accountwhen designing can give the user a product that he or she finds more attractiveand carries some emotional values too, rater than only addressing functionalneeds.
3.2 Driving
One of the larger studies that has been conducted on driving is the “100-carnaturalistic study” recording drivers and their actions immediately before ac-cidents and near accidents, sponsored by the National Highway Traffic SafetyAdministration in USA[Dingus et al., 2006]. The study took place in North-ern Virginian and the Washington DC metropolitan area and it was the firstinstrumented-vehicle study undertaken with purpose of collecting naturalisticdata in a a large scale. During the study no special instructions were givento drivers and instrumentation installed was unobtrusive. Data was collectedduring approximately 12 months for each vehicle and that resulted in 43 000hours of video and 2 000 000 miles of vehicle miles. 241 drivers and secondary
21
3.2. DRIVING CHAPTER 3. THEORY
Figure 3.5: Percentage of events for attention by severity level[Dingus et al., 2006, Figure RO.13, p. xli ]
drivers were a part of the study and events such as crashes, near crashes andincidents were recorded and analysed[Dingus et al., 2006, p. i].
3.2.1 Driving and inattention
The “100-car naturalistic study” recorded the effects of inattention to the for-ward roadway relative to incidents (crash, near-crash, etc., (see Figure 3.5) andtheir severity. 78% of the crashes and 65% of the near-crashes were causedby driver looking away from the road way just prior to the onset conflict[Dingus et al., 2006]. The inattention was classified in one of the four inat-tention categories combined [Dingus et al., 2006, p. xlii]:
• Secondary task i.e. using wireless devices (primarily cellphones), internaldistractions and passenger related secondary tasks (primarily conversa-tions)
• Driving-related Inattention to forward roadway e.g. checking the speedo-meter, rear-view mirrors or blind spots
• Drowsiness
• Non specific eye glance i.e. the driver briefly glances away form the road-way but at no discernible object or person
The most relevant contributor to crashes was inattention due to use of wirelessdevices (see Figure 3.6 on the next page)[Dingus et al., 2006, p. xli].
22
CHAPTER 3. THEORY 3.3. MODALITIES
Figure 3.6: Frequency of Occurrences of Crashes and NearCrashed[Dingus et al., 2006, Figure RO.14, p. xlii]
3.3 Modalities
All five sense are used to collect information but the visual, audible and tactileare the most common utilised in HMI for vehicles. When people use the IVIS orother functions in the car they typically have to look away from the roadway toperceive what they are doing when interacting with the system (see Figure 3.7on the following page). In the context of driving, focus on the forward roadwayand minimising eyes-off the road time is important for safety reasons and thus,a good system shall be efficient with regard to that (see 3.2.1 on the precedingpage).
3.3.1 Visual
Humans thrust their eyes more than any other and sense, an example of this iswhen you sit on a train and see another train depart. Even thought you don’tperceive any other tactile or audible input it feels like you move [Danielsson, 2001].
When people begin to look at an object they first rotate their eyeballs to-wards it (up to 275° per second), if the object is more than 15° from their currentline of sight their head moves 50 milliseconds later. Practically this means thatany object within 15° of the point of fixation can be fixated relatively quickly,since no head movement is needed [Olson et al., p. 79].
The vision is divided into three fields with decreasing fidelity and differentproperties, the Central Vision Field (CVF), the Functional Vision Field (FVF)and the Peripheral Vision Field (PVF) (see Figure 3.8 on page 25).
The CVF is a narrow field of vision where the eye focus and objects areperceived with clarity. Outside the CVF, the FVF forms a cone, 10-20° widewhere information is perceived with less clarity and detail. The rest of the fieldof vision is covered by the PVF that allows humans to perceive movement andlight but no colours or details [Bergmark, Gustafsson, 2008].
The ability of the lens in the eye to change refractive power is called ocular ac-commodation, and it occurs when objects are viewed at a closer distance than sixmeters, that is, at a closer range than optical infinity, when light rays stops en-tering the eye in parallel [Pavan-Langston, 2007, Bergmark, Gustafsson, 2008].
23
3.4. DISPLAY POSITION CHAPTER 3. THEORY
0.8
2 3 4 5 6 7
1.0
1.2
1.4
1.6
1.8
speed vent
following traffic
casette tapeinfo lights
balance/volume
temperaturedefrost/fuel economy
fuel range
fan
destination distance
destinationdirection
correct directionzoom level
cross street
road name
road distance
tune radio
power mirror
cruise control
heading
time
tone
fuel remaining
Number of Glances
Nav
igat
ion fea
ture
sA
vg. G
lance
Len
gth, [
s]
1.4
1.6
1.8
destinationdestination distance distance
destinationdirection
correct directionrect directionzoom levelzoom levelzoom lezoom lezoom le
cross streetcross street
road nameroad name
road distanceroad distance
headingheadingNav
igat
ion fea
ture
s
Figure 3.7: Glance data
Infants posses great powers of accommodation but this power decreases withageing. At the age of 40 years a substantial amount of accommodation powerhas been lost [Pavan-Langston, 2007].
These features of the eye plays a role in the to two factors that easily can bevaried when designing a car. The display distance in degrees from the forwardfield of vision and the distance to the screen in measured from the eye, playsa role in the design of the car interior since they affect the transition timebetween the road and the IVIS display. For example, the further away thedisplay is mounted (in degrees) the longer the transition time. This issue willbe discussed further in 3.4.
3.4 Display position
Research shows that display position has an impact on driver efficiency, if thedisplay is placed far from the normal driving forward field of view, the driver’speripheral vision cannot be effectively used to detect unexpected events in frontof the vehicle [Rydstrom et al., 2008, p. 1]. The display position, measuredhorizontally from the true horizon (0º) in the drivers field of view, is one of thefactors that determines how far the driver has to look away from the road whenusing the IVIS. The transition time increases as the driver has to move the gazea longer distance. Ideally the screen should be positioned within 15º measuredfrom the true horizon of the forward field of view, since this gives the shortesttransition time (see 3.3.1 on the preceding page). Depending on the control type
24
CHAPTER 3. THEORY 3.4. DISPLAY POSITION
CVF
FVF
PVF
Point of fixation
FVF20 degrees
CVF
PVF 180 degrees
Figure 3.8: Different fields of vision [Bergmark, Gustafsson, 2008]
25
3.4. DISPLAY POSITION CHAPTER 3. THEORY
Figure 3.9: Display position in degrees [Bergmark, Gustafsson, 2008]
chosen for the IVIS, the screen is mounted in a high (above 22º) or low position(below 22º) (see Figure 3.9).
3.4.1 Low display position
The preferred mounting position for direct interfaces is in a low position, becausethe driver has to reach the screen to operate it (see Figure 3.12b on page 29).Focusing the display requires from the driver to look down and away from theroad and the eyes have to readjust focus from infinity to close range which takesome time [Rydstrom et al., 2008].
In a simulator study conducted by Volvo comparing a touch screen with arotary controller, it was found that the swerving on the road and the steeringwheel activity increased, when interacting with the touch screen, as comparedto the rotary controller. One interpretation of the results might be that thelower position of the touch screen (compared to the rotary controller) preventsthe driver from effectively using the peripheral vision to support the driving[Rydstrom et al., 2008].
+ Easy to reach for touch display purposes- The driver has to look away from the forward field of view
Table 3.1: Low display summary
3.4.2 High display position
The preferred mounting position for indirect interfaces is a high position, be-cause of the advantages of separating the control from the display (see Fig-ure 3.11 on page 29). A high display position has a smaller distance in degreesfrom the forward field of view, and less implications on refocusing the eyes com-pared to a lower display position [Olson et al.]. Experiments conducted with adisplay mounted in a high position showed lower lane deviation compared to
26
CHAPTER 3. THEORY 3.5. CONTROLS
low display position [Rydstrom et al., 2008]. A high display position requiresthe user to map the remote control to the display.
+ Does not require the driver to look far away from the forward field of view- Mapping can be harder than for direct interfaces
Table 3.2: High display summary
3.5 Controls
There is a tendency in the car industry to embed in-vehicle comfort functionsin multifunctional, screen based, control interfaces [Rydstrom et. al 2007]. Thecontrols to these interfaces are either direct or indirect and represents the auto-motive industry approach to centralised driver interaction. The different ap-proaches carries different advantages and disadvantages that will be discussedbelow.
3.5.1 Mapping
The term mapping describes the relationship between a control, the thing itaffects, and the intended result (see Figures 3.10a on the following page and3.10b)[Cooper, Reimann, 2003]. Its function is to reduce the need for any in-formation from a user’s memory to perform a task, and produce desirable be-haviour without the need for trial-and-error. Poor mapping is evident when acontrol does not relate visually or symbolically with the object it affects, requir-ing the user to stop and think, ”what’s going to happen when I turn this knob?”[Cooper, Reimann, 2003].
The concept of mapping is the relation between the mental model in theuser’s head and the objects conceptual model for how it can be used, whenthese coincide, then there is a close mapping. The designer usually expects theuser’s model to follow the designer’s mental model, but this is obviously notalways the case [Norman, 1988].
Mapping the graphics displayed on the screen in the car to the physicalcontrol is important to usability and many problems are found in this area inexisting systems [Rydstrom et al. 2005]
3.5.2 Indirect interfaces
Indirect interfaces are controlled via a device that is separated from the screen(see Figure 3.11 on page 29). A typical example of this interaction methodis the BMW iDrive. This type of control is a bit less intuitive than directinterfaces, since the user has to map the interface on the screen to the inputdevice. However, it is preferred by some car manufacturers because the screencan be mounted closer to the windshield which gives a higher display position(see 3.4.2 on the facing page). Rydstrom and Brostrom found that a rotarycontroller (indirect interface) gave better performance in precision tasks such astuning the radio or zoom in on a map, compared to touch screen, in a studyconducted in a simulator at Volvo [Rydstrom et al., 2008]. Another study thatinvestigated the usability of existing systems, found that for naive users, a visual
27
3.5. CONTROLS CHAPTER 3. THEORY
(a) Example of bad mapping, stove top
(b) Example of good mapping, power seatswitch
Figure 3.10: Example of mapping
touch screen seems to be the easiest to operate, while an integrated rotary switchseems to be the most difficult one to operate [Rydstrom et al. 2005]. Interestingand important to note was also that many usability problems was found to havelot to do with the graphical interaction solutions, and less to do with the actualmanual interaction principle [Rydstrom et al. 2005].
3.5.3 Direct interfaces
Direct interfaces have controls in direct proximity of the screen (see Figure 3.12bon the next page). A typical example of this is a touch screen, but it can alsobe a screen with buttons mounted very close to the screen, for example directlybeneath it, which is popular approach on mobile phones. This interface type re-quires the screen to be mounted within the drivers reach for the driver to operateit, therefore it is less flexible regarding display position than an indirect interface.Rydstrom and Brostrom reports from a simulator study comparing touch screen(direct interface) with a rotary controller (indirect interface) a higher lane de-viation for the touch screen [Rydstrom et al., 2008]. Contrary the touch screengave lower task completion times and less glances for input tasks, for exampleentering a destination or dialling a phone number [Rydstrom et al., 2008].
28
CHAPTER 3. THEORY 3.5. CONTROLS
(a) BMW iDrive
(b) Audi MMI
Figure 3.11: Examples of indirect interfaces
(a) Toyota touch screen
(b) Jaguar touch screen
Figure 3.12: Examples of direct interfaces
29
3.6. INPUT CONTROLS CHAPTER 3. THEORY
Figure 3.13: Example of touchpad with two buttons and static scroll bars
3.6 Input controls
3.6.1 Touchpad
A touchpad is pointing device that translates the motion and position of auser’s finger to a position on the screen (see Figure 3.13). Most touchpadsoperate by measuring capacitance between sensors in the pad, but some measureconductivity instead. The capacitive type can distinguish a finger but may haveproblems if it is dirty or in a glove, which are problems that the conductive typedoes not have. These different properties are a result of the touchpads physicalconstruction and indistinguishable to the user.
Another interesting feature of the touchpad is its adaptability, in the sensethat it can detect one or many fingers on the touch surface, and that softwaresignal processing takes place to interpret the input. Since the interpretation issoftware based, the behaviour and output from the touchpad can be customiseddepending on application and that makes it extremely flexible. This formlessnessmakes it possible for example, to simulate a dial, switch, button or any othertype of control represented on the display and enable the user to act accordingly.
Since the touchpad usually is a flat surface it lacks feedback in some ways,the surface can be grooved or equipped with a pattern that the user can feelwith the fingertips, but that is a static solution and thus not optimal for afuture proof product. However, techniques are available that directly stimulatesreceptors in the fingers and hands to produce a virtual texture, that can beadapted to the graphic representation of the interface[Senseg, 2009].
One distinct feature of the touchpad is the possibility interpret charactersdrawn on it with a finger. This can be particularly useful for languages usinglogographic (e.g. Chinese) or syllabic (e.g. Japanese) writing, while still beinguseful for alphabets.
A touchpad can be used with absolute or relative pointing style dependingon application and what feeling one want to convey. Absolute movement makesa direct mapping of the interface to the touchpad surface, and requires the userto press exactly the right spot on the pad, to hit a specific area in the interface.Relative movement starts from a selected element and the selection is movedrelative to the current marked element (see Fig. 3.14 on the next page).
30
CHAPTER 3. THEORY 3.6. INPUT CONTROLS
Figure 3.14: Relative and absolute selection style on touchpad
+ Function and form can be defined by software and graphics+ Can handle gloves and wet weather with the right touchpad technology+ Multi-touch gives access to more dimensions that x- and y only+ Sign input with fingertip- Formless, the interface need to be represented on the screen- No intrinsic feedback
Table 3.3: Touchpad summary
3.6.2 Touch screen
A touch screen is a device that can detect presence and location of a touch withinthe display area, generally one or more fingers, but sometimes also a stylus (seeFigure 3.15 on the following page). The touch screen enables one to directlyinteract with the content on the screen, and is easy to learn for a naive user,since it does not require any new interaction strategy [Rydstrom et al. 2005].Since the interface is defined on the screen it is very adaptable and flexible andcan be made to fit whatever the developer want. One potential drawback is thatthe screen has to be mounted in a relatively low position to make sure that thedriver reaches the screen while driving, which forces the driver look away fromthe road. Haptic feedback is not usually not present, however, techniques areavailable that directly stimulates receptors in the fingers and hands to producea virtual texture on the screen [Senseg, 2009].
+ Direct manipulation makes it easy to use+ Mapping is good+ Adaptable+ Typically liked by users- Requires a low display position- Little feedback presently
Table 3.4: Touch screen summary
31
3.6. INPUT CONTROLS CHAPTER 3. THEORY
Figure 3.15: Example of touch screen displaying a user interface
Figure 3.16: Example of rotary switch (BMW iDrive)
3.6.3 Rotary switch
The rotary switch (see Figure 3.16) is a device that can be rotated or pushedin different directions to actuate functions. It is usually are located on thecentre console in the car to create an indirect interface (see 3.5.2 on page 27),where the system display has a high location on the dashboard. The rotaryswitch can be moved in four to eight directions for menu selection, rotated formenu navigation and pressed or pulled for activation. Rotary switches can beequipped with programmable haptic feedback, which makes it possible to utilisethe touch sense instead of only the visual modality, to minimise eyes of the roadtime when driving.
Some drawbacks with a rotary controller is mapping and character input.For example, mapping a rotational manoeuvre to a linear vertical movementcan be a bit unintuitive and a Chinese characters requires some effort.
+ Combines buttons and dials+ Remote controlled interface enables high screen position+ Haptics enables eyes off the screen use- Locked by physical constraints- Mapping can be a bit hard
Table 3.5: Rotary switch summary
32
CHAPTER 3. THEORY 3.6. INPUT CONTROLS
Figure 3.17: Example of buttons
3.6.4 Buttons
When thinking of buttons one of the more fundamental properties their physicalform which can be shaped to give affordances and hint function. Basic proper-ties of buttons in this sense are tangibility, physical representation and directaccess. They are unsophisticated, they can have a form and texture that givesgood affordance. However, they are often static and uninteractive and in largenumbers they can create a visual clutter. Nevertheless they are very useful andefficient for many applications.
+ Direct manipulation+ Simple and intuitive+ Tangibility- Static- Uninteractive
Table 3.6: Button summary
3.6.5 Joystick
A joystick is an input device consisting of a stick that pivots on a base andreports its angle or direction to the device it is controlling. Joysticks can beequipped with haptic feedback to aid the user and easy load of the visual mod-ality. One advantage of the joystick is the mouse-like behaviour combined withthe tangibility and the physical rigidity that makes it intuitive and robust to use.A drawback might be limitations that it usually only has two axis to operateon and not push or pull features such as the rotary switch.
+ Tangible and rigid+ Haptic feedback- No extra dimensions e.g. push or pull
Table 3.7: Joystick summary
33
3.7. MENUS CHAPTER 3. THEORY
Figure 3.18: Example of joystick (Lexus Remote Touch)
3.7 Menus
3.7.1 Menu design
Menus can be structured in an a number of ways, but the most common approachis to have some sort of system that is intuitive in relation to the context. Forexample, positionally and historically arranged systems, each have advantagesand disadvantages depending on application. One study investigating menustructure and performance comes to the conclusion that positionally arrangedmenus are fast and easy to remember after some use if the content does notchange over time[Somberg, 1987].
The same study concluded that historically based menus second by alpha-betically arranged menues are very fast too, and a good choice depending onapplication. For example, alphabetical menus would be efficient to find a namein a list, and a historically based list would be useful for the last called numbersdialled [Somberg, 1987]. This leads to the conclusion that there should be somebenefits in using a constant menu for the high-level choices, and other arrange-ments for lower levels such as alphabetic list in a phone book, in an infotainmentsystem.
3.7.2 Menu interaction
When scrolling in menus acceleration can be used to faster reach a certainposition. A study designed to test this with a scrollbar on a computer, foundout that a scrollbar with a three piece slider handle (fast, medium, slow) wasthe fastest for selecting an element in an alphabetically sorted list in an initialstudy[Ahlberg et al., 1998, p. 7]. A redesign was purposed with two differentspeeds for the position based slider, instead of three, to better fit user patterns.The study concluded that more research for touchpad and touch screen areneeded to draw any valid results, since they specifically did not test them asinteraction devices [Ahlberg et al., 1998, p. 7]. However, the positive resultsrendered when using acceleration in menus indicates that it might be worthinvestigating with a touchpad as interaction device.
34
CHAPTER 3. THEORY 3.8. TEXT INPUT
Figure 3.19: Example of pie menu [Kurtenbach, 1993]
3.7.3 Alternative menus
There are a number of ways to arrange menus and interaction in them and oneparticular example of this is the pie menu (see Figure 3.19). A pie menu is aformat where the items are placed along the circumference of a circle at equalradial distances from the centre [Callahan, 1988]. Pie menus gain over tradi-tional linear menus by reducing target seek time, lowering error rates by fixingthe distance factor and increasing the target size in Fitts’s Law, minimising thedrift distance after target selection, and are, in general, subjectively equivalentto the linear style [Callahan, 1988]. Experiments conducted have shown that 4,6, 8 and 12 item menus and on-axis items enhance performance, compared toa linear menu, when using a stylus[Kurtenbach, 1993]. This kind of menu canbe space consuming in terms of display area, and it does not perform as wellwhen using an odd number of menu items, which might be a limitation for someapplications[Kurtenbach, 1993].
3.8 Text input
3.8.1 1- and 2-dimensional keyboards
An on-screen keyboard can be represented in one or two dimension and havedifferent properties depending on input method (see Figure 3.20 on the followingpage). For example a linear keyboard would be a good method to input textwith a rotary device, since the rotating motion can be mapped to the movementof the cursor that selects the letter to input.
A 2-dimensional keyboard can be utilised with input methods that give ac-cess to two axes of movement, such as a touchpad or touch screen. Differentcombinations of keyboard layouts exist, QWERTY, DVORAK, alphabetic, aswell as keyboards with groups of keys, but no focus will be put on keyboardlayouts in this report.
3.8.2 Handwriting recognition
Recognition of characters can be done, for example with an input from a touch-pad or a touch screen, and it can be an alternative to on-screen keyboards dueto a number of reasons. Firstly, improvements in driving performance and inputefficiency has been observed in tests when utilising handwriting over on-screen
35
3.9. MULTI-FUNCTIONAL CONTROL PLACEMENTCHAPTER 3. THEORY
Figure 3.20: Example of 1- and 2-dimensional keyboards
keyboards [Burnett et al. 2005]. Secondly, languages using logographic (e.g.Chinese) or syllabic (e.g. Japanese) writing benefit from handwriting recogni-tion, while still being useful for alphabets. One disadvantage with recognitionof handwriting is the lack of support the user get from visually seeing the key-board on-screen, but that could probably be remedied with proper interfacedesign that supports the user.
3.9 Multi-functional control placement
The consolidation of distributed controls to one powerful centralised multi-functional control raises the bar for the system design. A badly designed cent-ralised control will have larger implications on the system behaviour, than forexample, if a a single function button was placed on a bad spot. The multi-functional control needs to be placed at a proper location within the car, in reachfor the driver and preferably a passenger too, in order to create a universallyusable system that also is safe to use.
3.9.1 Centre armrest
The standard place to mount multi-functional controls is in the centre armrestbehind the gear-stick. The driver and the passenger can easily reach this posi-tion. BMW iDrive, Mercedes COMMAND and Audi MMI uses this placementfor their controls. This position gave the lowest level of distraction in a touchpadclinic study conducted by Volvo HMI [Petterson, 2007]. The control is operatedwith the left or right hand depending on if driving takes place on the right-or left-hand side of the road. Since most people are right-handed this raisesan issue about of how dexterous the driver is the dominant and non-dominanthand, and if that causes any issues, which is discussed in 3.10 on the next page.
3.9.2 Door-side
Placing a multi-functional control in the door-side armrest would prevent pas-sengers from using it, and this placement has the same problems with the dom-inant hand as then centre placement.
36
CHAPTER 3. THEORY 3.10. DOMINANT HAND
3.9.3 Distributed
Possibly controls could be distributed in different locations in the car. Thiscould for example, be used for temperature control and volume control, but itis at the same time a counter-intuitive solution, since the approach to introducemulti-functional controls aims to reduce the number of controls scattered in thecar.
3.9.4 Steering wheel
A touchpad could be mounted on the steering wheel within reach of the thumbsor in the centre hub. One disadvantage would be that the pads would movewhen you make a turn, otherwise than if the hub is fixed. In a study conductedby Volvo the mounting position at the steering wheel gave slight longer taskcompletion times [Petterson, 2007]. Participants that evaluated both the steer-ing wheel position and centre armrest position found difficulties when steeringand giving input at the same time [Petterson, 2007]. In that test the touchpadwas mounted on the steering wheel hub and rotated with the steering wheel.
3.10 Dominant hand
An important question to address is the possible impact of dominant versusnon-dominant hand when using a multi-functional input control. Handedness isan attribute that defines unequal motor skills between left and right hand in anindividual. People are generally more dexterous in one hand, for example, if anindividual is right-handed, then that is the dominant hand. Kabbash, MacK-enzi and Buxton conducted an experiment comparing the performance of thedominant versus the non-dominant hand over a large range of tasks. A track-ball, mouse and a stylus was compared to elicit task performance and precisionbetween the different hands. The results indicated that for rough pointing ormotion, the non-dominant hand is as good as the dominant hand across a largerange of task difficulties [Kabbash et. al, 1993]. They also concluded that it isappropriate to use the non-dominant hand for tasks that do not require pre-cise action, such as scrolling and if the non-dominant hand is used for pointing,wide targets should be used [Kabbash et. al, 1993]. The different input devicesused showed some different trends during test; while there was the least changebetween hands with the trackball, non-dominant performance with the mousewas still far superior [Kabbash et. al, 1993].
37
3.10. DOMINANT HAND CHAPTER 3. THEORY
38
Chapter 4
Technology
An overview of the experimental setup will be presented in this chapter and atechnical explanation will be given of how the prototype was constructed.
Four basic parts were needed to build the concept:
• Capture program
• Touchpad
• Laptop
• Driving simulator or car
The capture programme handles gesture recognition and runs on the touchpad,which is an iPod touch. The touchpad sends the interpreted multi-touch gestureswireless as input to the simulation that runs on the laptop. The laptop isconnected to the screen in the simulator or car which displays the simulation(see Figure 4.1 for an overview of how the different components are connected).
4.1 Simulation
The simulation created during this project is an Adobe Flash application thatcan be run on any PC. It was written with an open-source framework (AdobeFLEX) in the programming language ActionScript 3 which generally is known asa Flash. It is a popular environment for building simulations because it is a fastand capable for interactive software prototyping with good support for graphics,video and interactivity. The simulation was run on an Apple Macbook and with
Figure 4.1: Overview of prototype setup
39
4.2. CAPTURE PROGRAM CHAPTER 4. TECHNOLOGY
Figure 4.2: Detailed overview of prototype setup
an Apple iPod Touch as input device which was rigged in the test vehicle (see 4.3)and in the driving simulator (see 4.4) for the user testing sessions.
4.2 Capture program
To capture the multi-touch events from the touchpad (iPod Touch) and sendthem to the laptop, a program called Jaduu VNC had to be installed on the iPod.Jaduu VNC can detect one-, two- and three finger gestures which translates topointer movement, scroll up and down, zoom and program switching on thelaptop. These gesture were remapped to fit the gestures thought up during theproject.
4.3 Test vehicle
Volvo Cars Corporation in Gothenburg has a specially equipped Volvo XC90 (seeFigure 4.3 on the next page) dedicated for use in clinics and user tests of driverenvironment and IVIS. All controls except those for driving can be connectedto a laptop, which makes it possible to use the controls in the car as inputs tothe simulation. The car looks and feels like any regular production car whendriven, but it has some extra features. In the middle of the dashboard, neatlyintegrated, close to the windshield, a display has been mounted. The displayis a bit larger than in a production car and it can be used to test graphics,interfaces and other visualisations under real driving conditions.
4.4 Driving simulator
The driving simulator at Lindholmen Science park is a so called fixed base sim-ulator that consists of the interior from a car, assembled in a room in front of
40
CHAPTER 4. TECHNOLOGY 4.5. TOUCHPAD
Figure 4.3: Volvo XC90
projector screen. It has a steering wheel, gear shifter, dashboard with instru-ments and a display, glove compartment, etc., (see Figure 4.5 on the followingpage) to mimic the interior of a car. The simulator is equipped with force feed-back that mimics mechanic resistance, to create a feeling that one is driving areal car. Thus, it works a bit like an advanced video game.
Additional equipment to measure eye movements and reaction times canbe mounted to make precise measurements, but none of that was used in thisproject, since it had required a substantial amount of time to setup and theprototype was unpolished. The experiments conducted in the simulator focusedon observing user behaviour and collecting general opinions instead.
The driving experience is a bit different from a real car, since there are noperipheral distractions and there is no real sense of speed either. Driving insimulators affect peoples judgement and make them overestimate their drivingcapabilities, compared to a real driving situation[Jamson, Merat, 2005]. Oneexplanation might be that no physical forces acts on the driver, which makesit is hard to estimate how hard one is breaking or how fast one is accelerating,etc., when driving in the simulator. However, this will not affect the outcome ofthe experiments conducted during the project, because user tests were focusedon user experience and not task performance.
4.5 Touchpad
During the first iteration an effort was made to find a multi-touch enabledtouchpad that could be connected to a laptop to run the concept simulationin a car. At first this seemed to be an easy task, but few companies producemulti-touch enabled touchpads and those who do, sell them almost exclusivelyto computer manufactures for integration in laptops. After a futile search fora touchpad with the right specifications it was decided to use an Apple iPodTouch (see Figure 4.6 on page 43) as a substitute instead. The iPod has multi-touch capabilities, and can wireless be connected to a laptop, with an auxiliaryprogram called Jaduu VNC [Jaduu VNC, 2009] . Jaduu VNC can forward themulti-touch gestures to the laptop that runs the simulation. Hence, renderinga modular multi-touchpad enabled laptop that can run the simulation.
41
4.5. TOUCHPAD CHAPTER 4. TECHNOLOGY
Figure 4.4: Lindholmen Open Arena driving simulator
Figure 4.5: Driving simulator interior
42
CHAPTER 4. TECHNOLOGY 4.5. TOUCHPAD
Figure 4.6: Apple iPod Touch
43
4.5. TOUCHPAD CHAPTER 4. TECHNOLOGY
44
Chapter 5
Realisation
The realisation was divided into three parts which each consisted of a numberof steps and a series of questions that needed to be answered. A brief overviewon the agenda is listed below:
• First iteration
– What is the purpose of the thesis?
– Who are the stakeholders and what are their initiatives?
– Create four concepts on paper
– Choose two concepts for further development
• Second iteration
– Implement a concept
– Test the concept in the simulator
– Observe users and gather opinions
• Third iteration
– Create a graphic identity
– Improve implementation after observations
– Conduct user testing with experts and regular users
– Identify further studies
5.1 First iteration
The purpose of the first iteration was to understand the problem and the dif-ferent stakeholders. What were the people involved in the project at Volvoexpecting, and what were the reasons to initiate the thesis? These people hadsomething in mind and expected touchpad to do something for them. How couldthis information be collected and understood? Why was it important? Intro-ductory literature was read to understand different input technologies and theiradvantages and disadvantages, what the IVIS should do, how it is evaluated,what are the limits and why are these limits imposed?
45
5.1. FIRST ITERATION CHAPTER 5. REALISATION
The goal of the first iteration was to understand the problem to be able tocreate an early sketch of a system on paper. That should be used to furtherdevelop the concept and implement it in the second iteration.
Could basic building blocks be identified that could serve as an interactionplatform that is complete, expandable and provides future development?
5.1.1 Knowledge gathering
Literature was reviewed in the beginning of the first iteration as well as through-out the whole project. Interviews were conducted with people at Volvo to get anunderstanding of the problem from their point of view. Three semi-structuredinterviews (see 2.5 on page 13) were conducted at Volvo Torslanda facilities withparticipants from Volvo Product Planning, Volvo Design and Volvo HMI.
During the first interviews information from different experts were collectedwith the objective to use their demands for using a touchpad was collectedfurther development. Summarised by stakeholders:
• Drivers
– Functions used often should be easy accessible
• Volvo Product Design
– Build a car that affects the user– Build a car that has iconic features
• Volvo Human Machine Interaction
– Build a system that is safe and easy to use– Usable by everyone, old or young, disabled or not, regardless of edu-
cation level
• Volvo Product Planning
– Build a system that is easy and intuitive to use– Support handwriting recognition– Customisation should be possible to different customer and market
segments
5.1.1.1 Drivers
No drivers were interviewed with explicit focus on driving and touchpad usage,since touchpad controlled systems are not present in cars today. Volvo has notpreviously conducted any studies with focus on what parts of the infotainmentsystem the drivers use, how they use it and why they use it [Moric, 2008].Considering 80/20 design (see 3.1.4 on page 20) focus should be placed on tasksthat are performed regularly and a presentation of these have been provided inthe research questions 1.8 on page 9.
It is interesting to notice that time we pass in a car is of value in other signific-ant ways beyond reductions of traditional economic models [Brown et al. 2008].This means that a car is not only a way of transport, but it carries some emo-tional values too which should be considered in a design:
46
CHAPTER 5. REALISATION 5.1. FIRST ITERATION
• Functions used often should be easily accessible
5.1.1.2 Volvo Product Design
A lot of effort has been invested in the visual trim and aesthetics of the car.Volvo Product Design see a possibility to integrate the touchpad technology ina way that will be beneficial to Volvo in the future. The IVIS interaction is acentral part of the car, within an area not driven or limited by nature forces, forexample, as the engine, but instead on an intellectual level, which makes thisarea susceptible to new radical concepts and ideas [Gordh, 2008]. The oppositeof this is e.g. the engine, it powers the car, but basically it is and energyconverter, more or less refined, thus limited by laws of nature.
There seems to be a vision that new IVIS interaction concept can become afeature, that make Volvo’s cars iconic, a bit like the traditional sign on the frontof the car. Thus something that defines Volvo as a company [Gordh, 2008]:
• Build a beautiful car that affects the user
• Build a car that has iconic features
5.1.1.3 Volvo Human-Machine-Interaction
The Human-Machine-Interaction (HMI) group evaluates and develops togetherwith other departments the interaction design in the cars. Their primary ob-jective is to build a car that is safe and easy to use. HMI are concerned whethertouchpad technology can compete with established technologies such as a rotarycontroller, and if any positive performance yield can be measured and in thatcase [Moric, 2008, Brostrom, 2009]:
• Build a system that is safe and easy to use
• Usable by everyone, old or young, disabled or not, regardless of educationlevel
5.1.1.4 Volvo Product Planning
The expectations from Volvo Product Planning (PPL) is that touchpad techno-logy can give Volvo a lead over competitors. PPL are very optimistic regardingtouchpad and they want a full featured experience that is expandable to handlefuture scenarios. Handwriting using the finger as a stylus is an important fea-ture for PPL, because of limitations currently imposed by international laws,that limits the use of other potential techniques. PPL want the system to beeasy and intuitive to use, something that can be expanded and customised, soldin packages and different categories of cars for different categories of customers[Volvo Strategic Planning, 2008]:
• Build a system that is easy and intuitive to use
• Support handwriting recognition
• Customisation should be possible to different customer and market seg-ments
47
5.1. FIRST ITERATION CHAPTER 5. REALISATION
• Multitask and context switching, e.g. answer phone and navigate, look upinformation on-line
• Surf Internet without any limitations
5.1.2 Brainstorm
A brainstorm session was conducted in the beginning of the first iteration atInteraktionsbyran to start up the project and get some of the participants formVolvo and Interaktionsbyran together to generate concept ideas. The brainstormwas conducted as described in section 2.6 on page 13 at Interaktionsbyran (2008-09-08). This was mostly aimed as an exercise in brainstorming and a primer toget some concept ideas to start exploring. A short introduction was held thatposed two questions as a warm up exercise.
• How can you interact with a touchpad?
• How can the interface be designed when using touchpad?
This proved to be a pretty good exercise and after a few minutes the team gotgoing with the idea generation. Some ideas were pretty wild e.g. using chords ona touchpad shaped like a guitar neck to control the IVIS. At first this might notseem like a good idea, but it turns out that it was interesting and usable, mainlybecause of two things. Firstly strumming a guitar is direct interaction withimmediate feedback and secondly the use of chords enables access to thousandsof combinations of tones in a very efficient way. On a touchpad and IVIS thiscorresponds to choosing between many options with direct access. A majordrawback is the learning threshold; its hard to learn finger positions for thechords on a guitar, but that did not stop the team from considering chords asmethod to directly access many different functions.
Another interesting discussions were how context, content and informationrelate to each other and what the implications of that are. The team imagineda system where:
• A contact from the telephone book could be dragged to the map and inthat case the map would display the persons location
• Same contact dragged onto a telephone icon would make a call to theperson
• Same contact dragged onto a messaging icon would send an email andSMS the person
These discussions gave the idea of ordering the menus by context, together withlogical grouping and redundancy, as it creates a system that is easily navigatedbecause it is logically structured, but it also fills the gaps where logical reasoningmight be a bit bungling. Research shows that menu width is preferred over depthby users without negative impact on performance [Landauer, Nachbar, 1985,Johansson, Walter, 2005]. An effort was made to create a logical grouping thatmade the interface as shallow as possible. A result of a wider hierarchy is ahigher information density in each screen which is preferred by users over lowerinformation density [Johansson, Walter, 2005].
48
CHAPTER 5. REALISATION 5.1. FIRST ITERATION
5.1.3 Checklist
The information gathered from the interviews (see 5.1.1 on page 46) togetherwith theory from chapter 3 on page 17 was combined to create a checklist, thatcould be used when weighing the concepts against each other, to determinewhich concept to be chosen for further development in the second iteration:
• Provide feedback of actions
• Permit context switching
• Provide good ergonomics
• Focus demand should be low
• Easy accessibility of information
• Supports multiple levels of information
• Be usable by a wide demographic
• Provide room for future extensions
• Natural interaction after learning
• Affect the user
• Have consistent navigation
• Have a good level of information density
• Be usable by an expert
• Be usable by a novice
• Be intuitive
5.1.4 Concepts
Here follows a description of the concepts that were created in the first iteration.
5.1.4.1 Concept 1
• Zoom information in layers
• This feature can be utilises when there is a need to browse dense inform-ation. This could for example be used to sort radio stations according togenre instead of selecting them from a list
• Settings accessed with buttons placed in appropriate positions in placeswhere they are needed
• Home button to access main menu and exit application
• 2-finger gesture for panning in menus
• 1-finger to move selection
• Pinch gesture to zoom
• Click works as an enter button
49
5.1. FIRST ITERATION CHAPTER 5. REALISATION
Figure 5.1: Concept 1
5.1.4.2 Concept 2
• Time machine style
• Gives a good overview between the different applications in the infotain-ment system
• Settings can be choose by panning sideways
• A click is required to select screen
• 3-finger gesture to access main menu
• 2-finger gesture for panning in menus including the time machine scroll
• 1-finger to move selection
• Pinch gesture to zoom
• Click works as an enter button
5.1.4.3 Concept 3
• Alt+tab style click and icon
• Gives a fast and unobtrusive way to choose application
• Settings accessed with buttons in the interface
50
CHAPTER 5. REALISATION 5.1. FIRST ITERATION
Navigation
Navigation
Media
Media
Climate
Climate
Navigation
Media
Media
Climate
Navigation
1. 2.
3. 4.
5.
Start view
Change view using two finger gesture
Media view selected Expands into full screen
Figure 5.2: Concept 2
51
5.1. FIRST ITERATION CHAPTER 5. REALISATION
Navigation
1. 2.
3. 4.
Start view Access change view using tap
Navigation
Switch between applicationsby dragging finger left or right
Navigation
Select view using tap
5.Media
Media view selected
Navigation
Figure 5.3: Concept 3
52
CHAPTER 5. REALISATION 5.1. FIRST ITERATION
*30% 20
1. Song2. Song3. Song4. Song5. Song
ArtistAlbum
Karl Johansson
o
2 km
4.2 km
2 bar2 bar
Contacts
*30% 20
1. Song2. Song3. Song4. Song5. Song
ArtistAlbum
Karl Johansson
o
2 km
4.2 km
2 bar2 bar
Start view Select group with finger
Enter selected group with tap Go back with home button
*30% 20
1. Song2. Song3. Song4. Song5. Song
ArtistAlbum
Karl Johansson
o
2 km
4.2 km
2 bar2 bar
Figure 5.4: Concept 4
• 3-finger gesture to access main menu
• 2-finger gesture for panning in menus
• 1-finger to move selection
• Pinch gesture to zoom
• Click works as an enter button
5.1.4.4 Concept 4
• Small multiples of information easy accessible
• Takes concept 1 and extends it to include small multiples of informationthat can be directly accessed
• 3-finger gesture to access main menu
• 2-finger gesture for panning in menus
• 1-finger to move selection
• Pinch gesture to zoom
• Click works as an enter button
53
5.1. FIRST ITERATION CHAPTER 5. REALISATION
Concept 1 Concept 2 Concept 3 Concept 4(Zoom) (Space) (Click) (Multiples)
Provides feedback of actions 3 3 2 3Permits context switching 2 3 2 3Provides good ergonomics 3 2 1 3
Focus demand (more points is less) 3 2 1 3Accessibility of information 2 3 2 3
Supports multiple levels of information 3 2 2 3Can be used by wide demographic 2 1 2 2
Provides room for future extensions 3 3 2 2Natural interaction after learning 3 3 2 3
User affected by interaction 3 2 1 2Consistent navigation 3 2 1 3Information density 2 3 2 3
Supports expert users 3 3 1 3Supports novice users 2 2 3 2
Intuitiveness 2 2 2 2Sum 39 36 26 40
Table 5.1: Comparison of concepts
5.1.5 Weighting
Weighting was used to assess the intrinsic development potential for each ofthe different concept sketches that were created during the first iteration. Thevalue represents a calculated guess of how much the concept can be evolved tofit the properties that were assembled from the checklist in 5.1.3 on page 49.The weight of the different properties was rated on a scale between 1 to 3, where3 is most in favour and 1 is least in favour of the selected property. The resultsfrom the weighing can be seen in Table 5.1
5.1.6 Discussion and conclusion
From the beginning some simple user testing was going to be performed toevaluate how users would handle gestures. Some simple paper prototypes werecreated that were supposed to be used in the test. The idea was to show twocards with mock up screens and four cards with possible gestures underneath,to test how people intuitive would act when interacting with the concept. Sur-prisingly it turned out that it was a pretty complex task to test and it requireda lot of effort to educate the test persons and introduce them to the idea behindthe concept. The decision was made in conjunction with the thesis supervisorat Volvo that a better and more strategic approach would be to create a simplemock-up prototype in flash that can be used in a focus group instead.
Using Windows or similar point-and-click based interfaces while driving isunsafe because it is demanding and complex for in-vehicle use. Buttons andother hit areas are small to make better use of the display space, and distractionsdoes not cause accidents when working with a computer at a desk. A solutionto create an interface that gives the driver more control, was to use Fitt’s lawas a foundation and create buttons with larger invisible hit areas, than made
54
CHAPTER 5. REALISATION 5.1. FIRST ITERATION
Figure 5.5: Gestures and increasing dimensions of freedom
the buttons easier to select and hit.The properties that was interesting to evaluate were if it would be possible to
educate the users to use three different gestures to navigate the system insteadof using just one finger. There seems to be many advantages with three gesturesto one finger navigation compared to the slight disadvantage that it will requiresome amount of training. That is, for each finger added two more dimension offreedom can be accessed (since the touchpad senses movements in 2-D), whilestill being backward compatible (see Figure 5.5). If the first level of interactionis selection, then all possible actions can be made accessed to one finger pointing.Using more than one finger will give direct access to a function such as pan orzoom which will make the touchpad more powerful. Of course this a gestureaction could be any arbitrary command, such as using two fingers to raise orlower the volume, etc.
The cursor should be hidden to minimise distraction and instead a markerthat shows which element is selected should be used. This transforms the touch-pad to a more dull control that should be easier and less demanding to use thana regular interface designed for a computer (see Figure 5.6 on the next page).
• Something is always selected on the screen to avoid that the driver getslost in the interface
• Buttons have large invisible hit areas to give the driver better control
• Relative selection style should be used
55
5.2. SECOND ITERATION CHAPTER 5. REALISATION
Graphic representation Corresponding hit areas
Figure 5.6: Selection model concept
• A physical back button that takes the driver back to the start view, willavoid confusion and guarantee general availability in demanding situations
• Back or erase buttons should be implemented in software and be availablein the different applications
The final concept chosen for development in iteration 2 was Concept 1, thatutilised zoom as a way to navigate the interface, it does not rank absolutelyhighest in the table (see Table 5.1 on page 54), but its closes competitors Concept4 would be more finicky to handle with many more small buttons. Concept 1 isalso more future proof since it would allow a number of new icons to be addedto the start screen, without any need for redesign.
5.2 Second iteration
During the second iteration a concept was implemented, to be used in the user-testing at the simulator at Lindholmen Science park. A field study was conduc-ted at a large electronics store called MediaMarkt in Hogsbo in Gothenburg anda test drive was conducted by the author in a real car. The second iterationfocused on implementing the result from the first and observe users handlingthe system to find weaknesses and areas to improve as well as collect commentsfrom users.
5.2.1 Store browsing
A field study was conducted at the electronics store MediaMarkt in Hogsbo tomake a hands-on approach testing available touch interfaces on different devices.Phones and navigators were the most common products utilising touch techno-logy in the stores inventory and these were chosen as the primary targets. Havinga large array of phone and navigators at hand to compare makes it very obviouswhich ones are a well designed and which are not. The most basic functionalityof the devices was tested rather quickly to experience how well they fulfilledtheir intended purpose, for example, a phone should have an easily accessiblephone book since its primary use is calling, likewise a GPS-navigator shouldpossess a method to enter a destination easy. Most of these products preformedmediocre in this test but one GPS-navigator intended for car use car stood outfrom the competition because of the following features:
56
CHAPTER 5. REALISATION 5.2. SECOND ITERATION
• Relevant and often used menu alternatives were easily accessed, that madethe interface fast to navigate and easy to overlook
• Large buttons with icons that properly resembled the intended function
• Logical and consistent grouping of buttons
• Arrows conveniently placed in the lower screen-corners that allowed undoand redo actions
• Wide menu hierarchy in favour of depth
• Consistent interaction
These design decisions made the unit easy and fast to handle and created apositive user experience.
5.2.2 Simulator test
The first simulator test took place at Lindholmen Science Park in the drivingsimulator (see 4.4 on page 40) that was booked for this particular event. Amixed group of 14 with different backgrounds were taking part in the study.The participants had been informed that they were going to attend a shortdemonstration and perform a small test of the prototype in its current state atthat time. The group consisted of 4 women and 10 men and their age rangedfrom 25 to 55. The simulator had been booked for the whole day and theparticipants arrived in small groups of 3-4 persons at intervals during the day.
The functions and gestures available in the concept were presented for theparticipants to make sure that all features were covered by the test. No specificscenario had been planned for the participants, but they were instructed toimprovise when testing the concept while driving at the same time. However,some care was taken to make sure that all participants covered all features ofthe concept.
The only driving environment in the simulator was an undemanding slowlywinding rural road with sparse traffic which created a very unchallenging drive.
Notes were taken on how the participants preformed and how they actedin the simulator, observed from a position from the left side of the simulator,with a good overview of the driver and their actions. No particular protocol wasfollowed during this study, because the objective was to get an understandingof how people related to the touchpad controlled interface.
5.2.2.1 Prototype
The prototype was pretty rough during this stage of the process. The graphicdesign did not fit in very well in the car interior, but the concept contained themost important parts that were going to be tested:
• Interface structure
• Gestures
• Home button implemented in software
• Buttons with large hit areas
57
5.2. SECOND ITERATION CHAPTER 5. REALISATION
• Relative mouse navigation
• Pointer style control
5.2.2.2 Result
• Setting the expectations for the participants is important for the test res-ults. The participants has to be in the right state of mind to properlyevaluate the prototype
• The zoom function is really appreciated and useful
• The pinch gesture for zoom is not optimal, it is a bit awkward becauseof the angle created between the hand and touchpad. The driver has tobend the wrist in an uncomfortable way since the touchpad is mounted ina low position behind the gear stick. It is better to use three fingers forzoom as this allows the driver to stretch the wrist. This can be concludedafter observing the participants try to use the touchpad mounted on theproposed position between the seats
• The texture of the touchpad is important for ergonomics, the surface hasto be rough enough for the fingers not to stick at the surface. This canbe concluded since the participants complained that the surface on thetouchpad was too sticky. By attaching a piece of paper on the touchpadthat problem could be solved, since it made the surface more rough. Testswith the participants confirms that this was preferred compared to theglass surface
• Positioning touchpad between the seats at the end of the armrest wasgenerally liked. One problem that might arise is that drivers of differentlength can’t optimally use the touchpad if it is fixed, because of their phys-ical differences. This can probably be solved by making a good armrestwith proper support for the wrist and palm of the hand
• The interface was finicky because of the pointer style navigation that wasavailable, this was exactly what the courser pointing style was going tofix, but this confirms the idea and the solution
• Instead of using a zoom motion between the different view states, fadeswere put in place to ease the transitions between the different view states.They were not noticed at all, and no one were confused about whathappened when they switched between view.
• Driving in the simulator and using the prototype felt pretty good over all
• There is a kind of finickiness using the touchpad compared to more physicalcontrols such as a rotary controller
• Wide sweeping movements was the preferred way to control an interfacein the car with a touchpad. Avoid small movements that are fragile andbrittle to give the driver control
58
CHAPTER 5. REALISATION 5.2. SECOND ITERATION
• It felt like haptics or audio feedback was missing when using the proto-type, it is very hard to get a feeling whats happening when using thetouchpad without these features because it lacks physical distinction. Forexample, a knob provides a fixed point in space that can be grabbed itand it feels rigid. This is a duality of constrained versus unconstrainedinteraction. There has to be some kind of boundaries that limits the userto a specific space. This is key in all interactions. That is in the same wayas a text-based interface is harder to use than an interface controlled witha mouse. The difference is a reduction in choice, from an infinite com-bination of words to a space confined within the borders of a 2-D screen,controlled with a cursor and buttons. This reduction of complexity is alsomirrored in the decreased power of expressing an action or command. Asan analogy, this is whats happening when using a touchpad in a car, withmore dimensions directly accessible there is a need to confine the space inwhich the users can move. This is exactly what the Fitt’s and Higg’s lawreduction that has been imposed on the touchpad interface does. It limitsthe available choices to a number optimised for each screen while mak-ing the areas that can be selected large to ease the selection and providecomfort and efficiency
• The participants complained that there was little distinctions between thedifferent views, for example it was hard to quickly distinguish which ofthe different views that was used, they wanted some kind of characteristic,such as an icon or colour theme, that could help them orient
• There needs to be s standardised way to answer simple questions like Yesor No
5.2.3 Graphic design
Test-participants in the simulator test (see 5.2.2 on page 57) commented on thegraphical design of the prototype that was used for testing. They thought itlacked car-ness and they wanted it to integrate better to give them a feeling of“car” instead of “computer”; two comments were - “It feels like Windows” and “Ithink it’s too square”.
A mood board (see Figure 5.7 on the next page) was put together, that wereintended to create a feeling of clean, glossy, fresh and car. The approach was toused to create very simple graphics that made the interface clear and functionalwith high contrast icons that distinctively communicated their function.
5.2.4 Driving test
The Volvo XC90 test vehicle (see 4.3 on page 40) was used for a driving test bythe author, to evaluate how the prototype handled in a real driving situation,compared to the more controlled environment in the simulator. The prototypewas in the same state that previously had been tested in the simulator andthe test consisted of a 30 minute driving session within the gates of Volvo’sTorslanda facilities. The driving conditions can be compared to driving in asmall city centre on a weekday with little traffic. There were some stretches of
59
5.2. SECOND ITERATION CHAPTER 5. REALISATION
Figure 5.7: Mood board for the graphic design
oncoming traffic, at speeds lower than 50 km/h and yielding to cars in somesituations. Compared to the simulator this environment was more demandingand that made the system harder to use.
The previous results from the simulator indicated that the system workedfairly good, but the driving test turned out to be really interesting, as the ex-perience differed a lot from the driving in the simulator. When driving a realcar the impressions are much more vivid and the sense of danger is more realcompared to the artificial danger in the simulator. It is easier to become over-confident in a simulator than in a real car, and overconfident driver behaviourhas been observed in simulator studies [Jamson, Merat, 2005]. Unfortunately,due to security regulations only the author could perform this testing, but itwould certainly had proven to be a very useful test if more participants hadbeen able to participate. However, the findings indicate that the final user-testing session ultimately should take place under real conditions, in the testvehicle, to get the best results, but unfortunately that will not be possible.
• Cursor speed is extremely important to give the user control over theinterface
• There is a tremendous leap from testing in a simulator to actually usingthe prototype in a vehicle, almost unbelievably large. After the drivingsessions it feels like something designed in the office is almost bound tofail
• If the system feels “ridiculously easy” to use in the simulator it is probablyjust “easy” or “not hard” to use under real driving conditions
60
CHAPTER 5. REALISATION 5.3. THIRD ITERATION
5.2.5 Discussion and conclusion
The driving test and the simulator test were probably the most important testsduring this iteration, which gave much information that would have been hard tocollect with other methods. When designing interfaces a lot of work takes placein a calm protective office environment, which a very different context comparedto driving. Experiencing the prototype while driving seems to be very importantin this particular field design, because the experience in the simulator and theoffice differs from reality.
The cursor speed has to be tweaked to make it usable when driving, to createa better system than the prototype, that was tested in this iteration
5.3 Third iteration
5.3.1 Graphic design
The graphics was revised in the last integration to create an interface that hadinfluences from Volvo’s concept cars to make it look a bit futuristic but also abit familiar.
5.3.2 The final concept
The main characteristics of the last prototype:
• Graphic design that blend in to the car interior
• Pointer is hidden and a coloured frame marks selection and make use ofFitt’s (see 3.1.1 on page 17) and Hick’s law (see 3.1.2 on page 18)
• Attempt to keep menu hierarchy shallow and wide
• Home button as a physical button (a la Wizard of Oz)
• Gestures implemented
– 1-finger
∗ Select∗ Handwriting recognition
– 2-finger
∗ Pan map in all directions∗ Scroll web pages up and down∗ Forward and backward in web history with a sweeping movement
including an elastic movement to indicate end or beginning ofhistory
∗ Choose between albums in a “cover flow” list of records
– 3-finger1
∗ Zoom in and out everywhere in the interface including maps,contacts, etc.
1This was actually performed with a pinch gesture on the touchpad, due to implementationsissues with the though up 3-finger gesture
61
5.3. THIRD ITERATION CHAPTER 5. REALISATION
5.3.3 Questionnaire
A questionnaire (see Appendix B on page 93) was created to be used for measur-ing user acceptance and evaluating the concept. The questionnaire was construc-ted of four parts; background questions, questions about touchpad proficiencyand touchpad preference, questions for open discussion among the participantsand questions grading the different parts of the concept. These four parts wereput together to get data from the test participants. The background part con-sisted of questions such as age, gender and education and the touchpad partwere aimed to get a measure on how found the participants were of using thetouchpad as input device generally. The questions for discussion were aimed toget the participants to act a bit like a focus group, discussing touchpad top-ics among each other, and the final part were questions where the participantscould grade the potential of the prototype.
The questionnaire was pilot tested and then rewritten to get the questionsmore straightforward so that the participants understood what they were askedto assess.
5.3.4 Evaluation
The goal with the final evaluation was to measure user acceptance and capturethoughts about what people think about using multi-touch interaction with atouchpad in an in-vehicle environment. The technical performance was not verygood in the concept so it was decided not to test and measure performance,since it doesn’t provide any significant value at this moment. Performancemeasures would be very important and interesting, but the numbers yieldedfrom the experiment would unfortunately not show any valid results, since thecontrols lagged behind a bit, which negatively contributed to the performance.However, the conceptual performance was very good, with many interesting con-cepts available (see 5.3.2 on the preceding page) that were the most importantparts to evaluate in this phase.
The participants in the evaluation consisted of two groups, one with expertsand one with novice and they had these characteristics:
• Expert group
– 9 participants
– Expertise in interaction design or a field related to car industry
– Education was on average a master degree
– Age ranging from 20-60 years old
• Novice group
– 7 participants
– No particular correlation in background e.g. one person was a teacher,one a fashion designer, one a project manager
– Education ranged from college to master degree
– Age ranging from 20-30 years old
The evaluation was divided in to five parts and took approximately 90 minutes:
62
CHAPTER 5. REALISATION 5.3. THIRD ITERATION
Figure 5.8: Concept average
1. Presentation of concept with screen-shots and explanations of the interac-tion
2. Questionnaire parts: Background and Touchpad
3. Concept evaluation in in-vehicle environment (no driving or driving sim-ulation)
4. Questions that served as subject for discussion among the participants
5. Questionnaire part: Prototype assessment
The results from this evaluation consisted of comments from the participantsand the data from the questionnaire (see Appendix B on page 93). Figure 5.8shows the average for the different parts of the concept. The coefficient ofvariation has been added to display the quote of the standard deviation to themean.
Figure 5.12 on page 65 shows the correlation between the concept averageand the touchpad average for each participant in the evaluation. The conceptaverage was calculated from the prototype assessment answers from question-naire and the touchpad average was calculated from the touchpad assessmentanswers from the questionnaire.
The solid line in Figure 5.12 on page 65 shows that the slope of the conceptaverage (y-axis) is very gentle as a function of the touchpad average (x-axis),which means that on average all users like the concept regardless what they thinkof working with a touchpad. The score would be on average at least 3.69/5 andat most 4.16/5 if the user disliked touchpad or liked touchpad respectively, thatcorresponds to a score between 7 and 8 on scale from 1 to 10. The character inputmethod was disliked by novice users, probably because it was implemented witha relative pointing style that made it hard to use. To make the character inputwork well with the finger as stylus, the touchpad should operate with absolutepointing style (see 3.6.1 on page 30) as this is the natural way for humans to write
63
5.3. THIRD ITERATION CHAPTER 5. REALISATION
Figure 5.9: Concept evaluation: Coefficient of variation
Figure 5.10: Touchpad average
64
CHAPTER 5. REALISATION 5.3. THIRD ITERATION
Figure 5.11: Touchpad average: Coefficient of variation
Figure 5.12: Prototype appreciation average as function of touchpad average
65
5.3. THIRD ITERATION CHAPTER 5. REALISATION
with for example paper and pen. Another comment from novice users were thelack of support to spell words. Some of the novice participants explained thatthey used the keyboard (e.g. on a computer) as a tool to remember how to spellcertain words, and that the lack of this made it hard to recall the spelling. Thisproblem can be solved by adding a context aware spelling support, that showsword alternatives in a list, from which the user can choose an appropriate word.The experts liked the input method better that the novice, probably becausethey had previous knowledge of other input methods and saw an advantage inusing the finger as a stylus.
• The character input method was disliked by novice users but liked byexperts
• The selection method was the feature least liked by the experts
• Zoom within contacts seemed to be a feature that novices though wereunnecessary
• The home button was liked by everyone
5.3.5 Discussion and conclusion
The results indicate that on average the concept is generally liked by all parti-cipants regardless of touchpad experience or preference. The score would be onaverage at least 3.69/5 and at most 4.16/5 if the user disliked touchpad or likedtouchpad respectively, that corresponds to a score between 7 and 8 on scalefrom 1 to 10.
This is interesting to know because a concept that is disliked by users wouldprobably be less successful within industry since car manufacturers are depend-ant on producing and selling cars. If an important feature of the car such asthe IVIS gives the driver a bad experience it is less likely that they buy the car.It is also likely that a system with decent performance, that demonstrates goodpotential and appeals to the users can be tweaked to meet necessary securityand safety standards, by using statistical data on driver behaviour and functionuse.
The most important feature of the touchpad in this regard, is its inherentability to change resolution and sensitivity, in combination with the possibilityto access many degrees of freedom in a natural way.
The age segment of the novice participants does not represent the typicalnew car buyer in Sweden. Considering that the development time from start toactual product within the car industry is many years, the age group will be abit closer to the average car buyer age when a potential system hits the market.
The age of the novice participants could have been spread in a wider rangethat 20-30 years old to possibly get a more varied result. The implications oftheir comparatively low age, might be that they are more susceptible to newtechnology and therefore gave the prototype a better grade in the evaluation,than an older group would have done.
66
Chapter 6
Result
A number of conclusions can be summarised from the test and development ofthe IVIS system. The proposed touchpad controlled IVIS system consists of atouchpad (see Figure 6.1) mounted between the front seats behind the gear stick(see Figure 6.2 on the next page), a screen mounted in a high position close tothe windshield and a interface (see Figure 6.3 on the following page) and aninteraction model (see 6.1) to control everything.
6.1 Interaction model
6.1.1 Selection
Touchpad interaction can be built with variable sensitivity on the pointer move-ment which resulted in two different modes that can be used for selection. Thefirst mode, fine, is fast and fluent but requires some amount of precision whenchoosing an element (see Figure 6.4 on page 69).
The second mode, coarse, requires less precision, it is slower, and works morelike a switch, e.g. press the switch three times to move the selection three steps.The same applies here but instead of pushing, a sweeping movement is appliedon the touchpad (see Figure 6.5 on page 69).
Figure 6.1: Touchpad with click
67
6.1. INTERACTION MODEL CHAPTER 6. RESULT
Figure 6.2: Proposed and tested touchpad mounting position
Figure 6.3: Interface
68
CHAPTER 6. RESULT 6.1. INTERACTION MODEL
Figure 6.4: Selection in fine mode
Figure 6.5: Selection in coarse mode
69
6.1. INTERACTION MODEL CHAPTER 6. RESULT
Graphic representation Corresponding hit areas
Figure 6.6: Selection model
The pointer in the interface should be a large and easily perceived marker toindicate what element that currently is selected on the display (see Figure 6.6).Some element in the display should always be selected, to avoid loosing controlof where the pointer is. Dividing the screen into large areas and giving thebuttons large invisible hit areas makes them easy to hit and gives the driverbetter control than if small buttons were used.
6.1.2 Gestures
Gesture give direct access to more dimensions of freedom than if only one fingeris used to control the system (see Figure 6.7 on the next page). This gives largebenefits at no cost since the interface can be designed to incorporate buttons forzoom and pan, etc., which makes it usable without gestures. Physical buttonscould be used to accommodate the same functions as the gestures provide, butthat does not give as fluid and natural interaction as the proposed gestureswould. In addition the more physical buttons added the less adaptable thesystem would be to future innovation, since those are built hardware and cannotbe upgraded or replaced without physical interaction, i.e. the car has to be takento a workshop and a mechanic has to fix it.
The gestures available in the final concept are:
• 1-finger
– Select– Handwriting recognition
• 2-finger
– Pan map in all directions– Scroll web pages up and down– Forward and backward in web history with a sweeping movement
including an elastic movement to indicate end or beginning of history– Choose between albums in a “cover flow” list of records
• 3-finger1
1This was actually performed with a pinch gesture on the touchpad, due to implementationissues with the though up 3-finger gesture
70
CHAPTER 6. RESULT 6.1. INTERACTION MODEL
Figure 6.7: Gestures in the final concept
– Zoom in and out everywhere in the interface including maps, con-tacts, etc.
6.1.3 Enter
There has to be a function in the touchpad to facilitate the Enter function foundon a computer keyboard. This should be a switch placed under the touchpadthat the driver can activate by pushing the touchpad downward (see Figure 6.1on page 67). It should be placed under the touchpad, to avoid being activatedby mistake. That would otherwise occur if the enter function would have beenactivated by a simply touching the surface of the pad as on a laptop.
6.1.4 Home button
A home button (see Figure 6.8 on the next page) in front of the touchpad takesthe driver to the home view (see 6.2.1 on page 74). This is provided to avoidgetting lost in the interface and always have a safe and comfortable way to getback to a secure position. This button can be used if the driver want to regaincontrol in a demanding situation or if the driver want to abort a task at somepoint. Different functions using short- and long-button pushes on the samebutton should not be used in the interface, as this can confuse novice drivers,which has been observed in clinics at Volvo[Petterson, 2009].
71
6.2. INTERFACE CHAPTER 6. RESULT
Figure 6.8: Touchpad and home button
6.1.5 Character input
The character input is accessed trough clicking in a text field and pops up as afull-screen overlay (see figure 6.9 on the next page). When entering charactersthe touchpad should be in absolute pointing mode to make it easier for the driverto input text. Context aware spelling suggestions should also be provided in alist, from which the driver can choose a suitable word, to support the driver,save time and make the system easier to use. Pop-ups can be very disturbing,and a better way to handle character input would be if the interface zoomedin on the selected text field, and handled the handwriting directly in the fieldinstead, complete with all help functions as described above.
6.1.6 Simple questions
When the IVIS system requires information from the driver a dialogue can bedisplayed that shows a few standard answers (see Figure 6.10 on the facingpage). These should be very simple to answer, with a quick sweeping motion bythe hand (regardless of the number of fingers), and be manageable in demandingsituations and have a consistent form and function. Question such as Yes andNo should be answered by a sweeping motion to the right and left respectively,that also corresponds to the back and forward function in the Internet view.
6.2 Interface
The ideas behind the interface will be presented here. One of the main conceptsis to make the interface more compact by reduce the number of choices. Forexample, it should not be possible to choose between CD/DVD/MP3 as sourcesof music. The music from all sources should be available and collected at one
72
CHAPTER 6. RESULT 6.2. INTERFACE
Figure 6.9: Character input dialogue
Figure 6.10: Simple questions
73
6.2. INTERFACE CHAPTER 6. RESULT
point - Music. There is no need for the driver to choose between these differ-ent sources if the interface presents the information in a proper and consistentmanner and this was seen as a major point to improve in the interface design.
6.2.1 Home view
The interface start in the home view (see Figure 6.11 on the next page). It isdivided into six different groups of functions with a large icon that resemblesthe functions and information collected under it.
The six categories where all functions has been gathered are:
• Contacts
• Navigation
• Communication
• Internet
• Entertainment
• Car settings
6.2.2 Contacts
Contacts contains a list of the drivers contacts and information associated withthem, such as phone numbers, addresses and email. The mix of information isa way to make the menu system more shallow and present more information onevery screen. The first view seen when starting the system is the Home view(see Figure 6.11 on the facing page). If the driver selects the Contacts icon andpresses the touchpad the contacts will be displayed (see Figure 6.12 on the nextpage). This view can be sorted like thumbnails or an alphabetic list dependingon the drivers preference (see Figure 6.13 on page 76). Selecting a contact andpressing the touchpad will take the driver to the contact where different types ofinformation can be accessed directly. For example, pressing the contacts home-address-button will create a route and start a route-guide to that point (seeFigure 6.14 on page 76). Likewise pressing the mobile phone number would callthe person.
6.2.3 Navigation
The navigation interface is very sparse and displays only a map in the concept(see Figure 6.15 on page 77). This is because the map scrolling 2-finger gesturewas tested and no focus was put on specific map functions. In a real implement-ation overlay panels could be added with search bars, destination entry, routedisplay and similar map related functions. Since the 2- and 3- finger gesturesare used to pan and zoom in the map, these functions are directly accessible onthe map in a natural way, while still being able to use 1-finger select for otherpurposes e.g. looking at points of interests, pressing buttons, etc.
74
CHAPTER 6. RESULT 6.2. INTERFACE
Figure 6.11: Home view
Figure 6.12: Contacts
75
6.2. INTERFACE CHAPTER 6. RESULT
Figure 6.13: Contacts as a list
Figure 6.14: Multiple types of information and actions available for a contact
76
CHAPTER 6. RESULT 6.2. INTERFACE
Figure 6.15: Navigation interface displaying a sample map
6.2.4 Communication
The communication view (see Figure 6.16 on the next page) contains means tosend either a text message or call someone with a phone bummer. There is noselection of receiver, SMS, MMS e-mail before the message is written becausethose are decisions that should be taken later, when the writing is done.
6.2.5 Entertainment
The entertainment interface should contain all sorts of entertainment such asmusic, video and radio which should be sorted under their respective category.For example, if the driver connects his portable media player to the car, the mu-sic from the device should be accessible under the Music category and the videosshould be accessible under the Video category. The interface (see Figure 6.17on the following page) contains only a sample of records that can be flickedtrough with a 2-finger gesture to test that way of interaction in an in-vehicleenvironment.
6.2.6 Internet
The Internet view contains a navigation bar where the driver can enter anaddress to go to. The pages can be scrolled up and down with a 2-finger gesturewhile back and forward in history are available by a 2-finger swipe left and rightrespectively. When the user hits the end of the history, the last page acts a bitlike it its attached to a spring and simulates resistance, to display that the thereare no more items in the browsing history.
Links are highlighted in blue as shown in Figure 6.18 on page 79 and a clickloads the linked page as in Figure 6.19 on page 79. For better accessibility thepages can be zoomed in as shown in Figure 6.20 on page 80 by using the 3-fingergesture. This enables unconstrained web browsing and universal access.
77
6.2. INTERFACE CHAPTER 6. RESULT
Figure 6.16: Communication view
Figure 6.17: Entertainment view displaying a number of record covers
78
CHAPTER 6. RESULT 6.2. INTERFACE
Figure 6.18: Internet view with dn.se loaded and the first article selected
Figure 6.19: Internet view with page transition in progress
79
6.3. CONTROLS CHAPTER 6. RESULT
Figure 6.20: Internet view with article zoomed in
6.2.7 Car settings
The car settings menu should contain all the different car properties that canbe tweaked or accessed. For example fuel consumption, oil change interval andinformation about the next service.
6.2.8 Favourites
The favourites bar on the right in the interface (see Figure 6.11 on page 75) workslike a permanent placeholder for shortcuts functions often accessed. For exampleif the driver wants to play a specific radio channel it can be made available here.Another scenario that could be addressed is when the driver want to listen toa playlist of favourite song, an Internet page with traffic information, a phonenumber, a waypoint to someones home, a home page with hours of operationfor a store, etc. The favourites bar is arranged after category to make it easy tofind a specific element:
• Arranged after category
• Provides customisation
• The drivers choices can be quickly accessed
6.3 Controls
The dimensions of the touchpad should be at least 11×8 cm, according to thetest results and have the same proportions as the display. The pad used theduring evaluation in this project was perceived to be too small with an area of8×5 cm. There shall be a large easily accessible volume knob in the car with apush function that mutes the speakers, and the climate and temperature con-trols should be separated from the IVIS system. Integration of these functions
80
CHAPTER 6. RESULT 6.3. CONTROLS
were perceived to be a bad decision by the test participants, since they are soimportant and frequently used.
The touchpad surface texture has to be a bit rough to decrease finger tosurface friction, which otherwise prevents the driver from controlling the system.
The touchpad should be mounted in an angle with the front edge placed a bitlower that the back edge. This gives a the user a consistent feeling of resistancewhen swiping the fingers in the back and forward direction of the pad. A goodarmrest and wrist support has to be in place behind the pad to give the userproper control of their finger-movements.
81
6.3. CONTROLS CHAPTER 6. RESULT
82
Chapter 7
Discussion and conclusion
Driving is a very physical experience with levels, buttons and forces acting onthe driver. A touchpad is not tangible in the same way as almost all controlsare in a car today and that is a challenge for using the touchpad as an inputdevice. Utilising a touchpad as interaction input device for an IVIS systemis definitely possible if the system has been designed with the advantages anddisadvantages of using a touchpad in mind. At first the touchpad appears to bea bit brittle compared to other techniques, mainly because of the lack of physicalproperties. This can be compensated for by making a more dull control in theinterface, which is possible since the sensitivity of a touchpad can be tweaked ina full spectrum from simple push button to high resolution control with multipledimensions available trough the use of gestures. Necessary haptic or auditoryfeedback can be added to give the system a more tangible impression and betterperformance. A study investigating the effects of bi- and tri-modal haptic andauditory feedback in a touchpad controlled IVIS system concludes that bi-modalfeedback was the most efficient, with auditory ranking slightly higher that haptic[Vilimeck, Zimmer, 2007]. However, this might be a result of the participantsbeing more used to auditory feedback and hence they responded better to that[Vilimeck, Zimmer, 2007].
• The use of a touchpad with multi-touch gestures gives access to manydegrees of freedom in a natural way of interaction. How many depends onthe number of gestures used. Three gestures have been proposed in thisthesis, which provides the basic interactions necessary for a complete IVISsystem including unrestricted Internet navigation
• The interface and the touchpad has to very responsive to be usable andfor the driver to remain in control
• The touchpad surface texture has to be a bit rough to decrease finger tosurface friction, which otherwise prevents the driver from controlling thesystem
• The touchpad should be mounted in an angle with the front edge placeda bit lower that the back edge. This gives a the user a consistent feelingof resistance when swiping the fingers in the back and forward directionof the pad. A good armrest and wrist support has to be in place behindthe pad to give the user proper control of their finger-movements
83
CHAPTER 7. DISCUSSION AND CONCLUSION
• Haptic or auditory feedback has to be added to create a fully functionalsystem in a real production car
• The touchpad is not as solid as a physical control which results in more fra-gile interaction, which can be compensated for by adjusting the resolutionand pointing style depending on application and driving situation
• The lack of multi-modal feedback affects the interaction and demands verygood graphic design to make the interface undemanding to use. A solutionto this is to use two modes of selection, one coarse driving mode that stripsthe system of functionality and allows the driver to perform basic actions,and one finer mode that allows detailed manipulation e.g. dragging andplacing a destination on the map when traffic situations allows that
• Reducing the amount of levels in the interface and provide different typesof information, relevant for the present context is a way to make the systemeasier and less demanding to use, while it also is appreciated by users
• The trend to put more functions into the IVIS system without investig-ating the user requirements and the actual use of these functions shouldbe considered to be revised. Very simple observation based studies can beperformed to gather information, which can be used as a starting pointfor this work towards a more user-centred design process.
• Statistical data about user patterns should be collected to better designthe system for the drivers. For example, mounting a small video camerain the ceiling of the car interior could be used to capture user activityand when the video is fast-forwarded the user patterns will emerge. Aspreviously discussed, this would be beneficial to the industry in terms ofproducing a better product, that in turn could help reduce the number oftraffic incidents
• The favourites bar proposed in this thesis works as a way to avoid com-plex menu navigation in a number of cases. This increases performancestatistically over time, since functions often used can be accessed withoutentering any menu at all. In combination with the home button this makesnavigation backwards in menus obsolete in the cases where a favourite isused
• Character input should be reduced to a minimum as it is resource consum-ing. Spelling support and word suggestions should be provided to simplifythe action
• The age of the novice participants could have been spread in a wider rangethat 20-30 years old to possibly get a more varied result. The implicationsof their comparatively low age, might be that they are more susceptibleto new technology and therefore gave the prototype a better grade in theevaluation, than an older group would have done
• The iterative process used during the course of this thesis has generallybeen useful and generated good results. The Youtube review methodgave sometimes alternate information that was hard to find at vendorshomepages, etc. For example, someone had filmed their first interaction
84
CHAPTER 7. DISCUSSION AND CONCLUSION7.1. FURTHER STUDIES
with the joystick controlled interface in a Lexus at a car fair. Among manythings it could be seen that it was a haptic joystick controlled pointer basedinterface which was hard to understand from other sources of information
• The driving test was a really good experience to understand the differencebetween a simulator and real world driving. Testing a concept in a realdriving situation early in the development gives information about whichare the most important areas to improve, which helps focus resources onthe most important problems
• The driving test performed in the real car gave a very good perspectiveon how different a real driving situation is from an office environment.It is recommended that system- and interaction designers try out theirprototypes in a real car during a prototype stage, to get an understandingof how one is affected by the system and what properties are important.The results are not trivial, for example, small animations can be verydisturbing when driving. Though they seem very subtle and informativeand well designed in the office. Another example is colours, they mightseem perfectly fine on a computer screen in the office, but in sunlight on adisplay made for a car they are too washed out to be useful. Yet anotherexample is the effect of input controls, what sensitivity they have, how youoperate them, etc. The control and the distractions you are exposed to,greatly affects the perception of a systems capabilities and weaknesses. Ina real driving situation these are perceived much different from an officeor simulator environment
In summary the results show that a rich multi-touch controlled interface can bedeveloped that users accept and like regardless of previous personal preferenceof touchpad usage, but it should to be equipped with feedback in one moremodality than visual to give the users proper control. Further areas to explorewill be discussed in the next section.
7.1 Further studies
To fully understand and elicit the performance of a touchpad controlled systemit would be feasible to build a prototype system with a low response time, as thisproved to be a very important feature to give the driver control of the IVIS. Alower response time would make it possible to get correct measurements of tasktime and mental workload to make a valid comparison with existing systems.The system does not need to have much menu items, but it should be somewhatrich and built on the principles presented in this thesis.
Another area that should be investigated further is the use of multi-modalfeedback, as this can be used to lessen the visual workload required to control thesystem. Existing technology can be used to create hapticons, brief programmedforces applied to a user through a haptic interface [Enriquez, MacLean, 2003].The feedback should be course and distinct to make it easy to recognise differentpatterns and functions in the system. An example of this could be a scenariowhere the surface of the pad creates a 2-dimensional matrix of virtual buttons(representing the main-screen). The drivers can feel the buttons with theirfingers and a distinct sequence of vibrations, that shakes the wrist support in aspecific sequence, for each unique menu element.
85
7.2. PERSONAL RECOMMENDATIONSCHAPTER 7. DISCUSSION AND CONCLUSION
• Build prototype with low response time to get performance measure
• Investigate if hapticons can make the system less demanding to use
7.2 Personal recommendations
The success of an IVIS system does not build upon the interaction method alone,as can be concluded from the examples in this report. In many cases the inputdevice works very well, but the drivers does not understand the graphics fromthe IVIS system, which makes the system less successful from a users pointof view. There are almost infinite combinations of controls, graphics, sets offeatures etc., and these components has to be considered as a whole to build agood, safe and useful IVIS system. It has to meet the basic needs of the driverbefore more features can be added, but since the application itself (driving) issafety-critical, care has to be taken no to stuff too much into the car as a whole.
The touchpad used in this thesis shows great potential for current and futureapplications because of its inherent flexibility. It can handle the full range ofinput, from the most primitive sweeping gesture to very fine mouse-pointer-likecontrol and by using multi-finger gestures its range can be extended to multipledimensions. If the interface is designed as proposed in this thesis, it would alsobe future proof and ready to handle other software applications than those pre-installed in the car when it is sold. A touchpad could in other words handlemost situations it would meet in the application as input device for an IVISsystem.
The major drawback of using a touchpad as input device is the lack offeedback compared to other car controls and a remedy to this is, technology thatcreates a feeling of touch in the users fingertips or a haptic device underneaththe wrist-support in the armrest. If the problem with feedback is addressed,then the touchpad technology could take a step forward as an IVIS control andbring a new type of interaction device into the car.
I think a prototype with a very responsive (in time) touchpad, in combina-tion with the proposed interface structure, bi-modal feedback and multi-touchinteraction model, would give good results compared to other systems. Espe-cially in areas where (relatively) unconstrained movement is feasible, such asweb-browsing or tasks involving panning, zoom or scrolling. It would also bean important step to make sure that the technology is safe to use in a carenvironment, as this could not be thoroughly tested in this thesis.
• Build a prototype with good response time and haptic feedback
• Gather data for statistical analysis of driver behaviour and IVIS usage.Design and optimise the system with these data in mind
86
Bibliography
[ABIresearch, 2008] Touch Screens in Mobile Devices to Deliver $5 BillionNext Year, [Accessed 2009-05-12],http://www.abiresearch.com/press/1231-Touch+Screens+in+Mobile+Devices+to+Deliver+$5+Billion+Next+Year
[Ahlberg et al., 1998] Ahlberg, C. & Schneiderman, B. (1998) The Alphaslider:A Compact and Rapid Selector, Department of Computer Science,Human-Computer Interaction Laboratory & Institute for SystemsResearch University of Maryland
[Benyon et. al., 2005] Benyon, D., Turner, P. & Turner, S. (2005) DesigningInteractive Systems: People, Activities, Contexts, Technologies,ISBN 0-321-11629-1
[Callahan, 1988] Callahan, J., Hopkins, D., Weiser, M., & Shneiderman, B.(1988) An empirical comparison of pie vs. linear menues, CHI ‘88,95-100, New York: ACM
[Cooper, Reimann, 2003] Cooper, A. & Reimann, R. (2003) About face 2.0,ISBN-10: 0764526413
[Bergmark, Gustafsson, 2008] Bergmark, M. & Gustafsson E. (2008) Develop-ment of a voice commanded HUD interface for an in-vehicle mobileapplication, Chalmers University of Technology
[Brostrom, 2009] Brostrom R. (2009), Volvo Cars Corporation HMI
[Brown et al. 2008] Brown, B., Laurier, E. & Lorimer, H. (2008) Driving and“passengering”: notes on the ordinary organization of car travel,University of Edinburgh, Mobilities, Volume 3, Issue 1 March 2008,pages 1 - 23
[Burnett et al. 2005] Burnett G.E, Lomas S.M., Mason B., Porter J.M. &Summerskill S.J. (2005) Writing and Driving: An assessment ofhandwriting recognition as a means of alphanumeric data entryin a driving context, School of Computer Science and Informa-tion Technology, University of Nottingham, UK, Department ofDesign and Technology, Loughborough University, UK, Advancesin Transportation Studies an International Journal 2005 SpecialIssue
87
BIBLIOGRAPHY BIBLIOGRAPHY
[Danielsson, 2001] Danielsson, M. (2001) Teknisk psykologi, Stockholm, Naturoch kultur
[Dingus et al., 2006] Dingus, T. A., Klauer, S.G., Neale, V. L., Petersen, A.,Lee, S. E., Sudweeks, J., Perez, M. A., Hankey, J., Ramsey, D.,Gupta, S., Bucher, C., Doerzaph, Z. R., & Jermeland, J. (2006)The 100-car naturalistic driving study, Phase II - results of the 100-car field experiment, Washington, DC: National Highway TrafficSafety Administration.
[Displaybank, 2009] In-depth Analysis: Touch Screen Panel Industry Trends[Accessed 2009-05-12]. Displaybankhttp://www.displaybank.com/eng/report/report.php?mode=show&id=698
[Engelbrektsson et al., 2004] Engelbrektsson, P. & Soderman, M. (2004) Theuse and perception of methods and product representations inproduct development: A survey of Swedish industry, Journal ofDesign Engineering
[Enriquez, MacLean, 2003] Enriquez, Mario J., MacLean & Karon E. (2003)The Haptic Editor: A Tool in Support of Haptic CommunicationResearch, Department of Computer Science University of BritishColombia, Sypmosium on Haptic Interfaces for Virtual Environ-ment and Teleoperator Systems (HAPTICS’03), IEEE
[Fitzmaurice et al., 2008] Fitzmaurice G., Matejka J., Khan A., Glueck M. &Kurtenbach G. (2008) PieCursor: Merging Pointing and CommandSelection for Rapid In-place Tool Switching, CHI 2008, April 5–10,2008, Florence, Italy, ACM 978-1-60558-011-1/08/04
[Gonzalez et al., 2007] Gonzalez Ivan E., Wobbrock Jacob O., Chau DuenHorng, Faulring Andrew and Brad A. Myers (2007) Eyes on theRoad, Hands on the Wheel: Thumb-based Interaction Techniquesfor Input on Steering Wheels, Human-Computer Interaction In-stitute School of Computer Science Carnegie Mellon University,Pittsburgh, PA 15213, Graphics Interface Conference 2007 28-30May, Montreal, Canada
[Gordh, 2008] Gordh Mikael, [2008-09-03], Volvo Cars Corporation Design,Gothenburg, Sweden
[Hsu, 2009] Hsu, A. [Accessed 2009-05-12]http://www.edn.com/article/CA6625439.html
[Jaduu VNC, 2009] http://www.jaaduvnc.com/ [Accessed 2009-03-15]
[Jamson, Merat, 2005] Jamson, A.H. & Merat, N. (2005) Surrogate in-vehicleinformation systems and driver behaviour: Effects of visual andcognitive load in simulated driving, Transportation Research F, 8(2). pp. 79-96. ISSN 1369-8478
[Johansson, Walter, 2005] Johansson, H. & Walter, K. (2005) InVehicle Screen-Density, Driver distraction and User Preferences for Low vs. HighScreen Density in intergrated Display, Lindkopings Universitet,Sweden
88
BIBLIOGRAPHY BIBLIOGRAPHY
[Jones, 1992] Jones, J.C. (1992) Design Methods, ISBN 0-471-28496-3
[Jordan, 2000] Jordan, P.W (2000) Designing Pleasurable Products, ISBN 0-415-29887-3
[Kabbash et. al, 1993] Kabbash, P., MacKenzie, I.S. & Buxton, W. (1993) Hu-man performance using computer input devices in the preferredand non-preferred hands. Proceedings of InterCHI ’93, p. 474-481
[Karlsson, 2007] Karlsson, M. (2007) Lyssna till kundens rost, Institutionen forprodukt- och produktionsutveckling, Chalmers tekniska hogskola
[Kelley, 2006] Kelley D. (2006) The ten faces of innovation, ISBN 1-86197-806-5
[Kurtenbach, 1993] Kurtenbach, G.P. (1993) The design and evaluation ofmarking menus, PhD. Thesis, University of Toronto, 1993
[Landauer, Nachbar, 1985] Landauer, T. K., Nachbar, D. W. (1985) SelectionFrom Alphabetic end Numeric Menu Treea Using A Touch Screen:Breadth, Depth, and Width, Bell Communications Research Mor-ristown, N.J.
[Lidwell et al., 2003] Lidwell, W., Holden, K. & Butler, J. (2003) UniversalPrinciples of Design, ISBN 1-59253-007-9
[Lovgren et al., 2004] Lovgren, J. & Stolterman, E. (2004) Design av informa-tionsteknik, ISBN 91-44-04203-5
[Moric, 2008] Moric, Azra (2008) Thesis supervisor, Volvo Cars CorporationHMI, Gothenburg
[Norman, 1988] Norman, D. (1988) The Psychology of everyday things, ISBN0465067093
[Olson et al.] Olson, P. L., Dewar, R. E., &Alexander, A. J. (2002) HumanFactors in Traffic Safety, ISBN 0913875473
[Pavan-Langston, 2007] Pavan-Langston, D. (2007) Manual of Ocular Diagnosisand Therapy, ISBN 0781765129
[Petterson, 2007] Petterson, I. (2007) Touch Pad: Usability, Distraction andAppeal Study, Volvo Cars Corporation HMI, Gothenburg
[Petterson, 2009] Petterson, I. (2009) Volvo Cars Corporation HMI, Gothen-burg
[Rydstrom et al., 2008] Rydstrom, A., Brostrom, R. & Bengtsson, P. (2008)Safety aspects of using multifunctional control interfaces whiledriving – a simulator study Department of Human Work Sciences,Lulea University of Technology, Sweden Volvo Car Corporation
[Rydstrom et. al 2007] Rydstrom, A. (2007) The effect of haptic feedback invisual-manual human-machine interaction, Lulea University ofTechnology Department of Human Work Sciences Division of En-gineering Psychology , ISBN: LTU-LIC– 07/04I– SE
89
BIBLIOGRAPHY BIBLIOGRAPHY
[Rydstrom et al. 2005] Rydstrom, A., Bengtsson, P., Grane, C., Brostrom, R.,Agardh, J., Nilsson, J., (2005) Multifunctional systems in vehicles:a usability evaluation, Proceedings of CybErg 2005. The Fourth In-ternational Cyberspace Conference on Ergonomics. Johannesburg:International Ergonomics Association Press
[Senseg, 2009] http://www.senseg.com [Accessed 2009-02-14]
[Somberg, 1987] Somberg, B. L. (1987) A comparison of rule-based and posi-tionally constant arrangements of computer menu items, Confer-ence on Human Factors in Computing Systems archive Proceed-ings of the SIGCHI/GI conference on Human factors in computingsystems and graphics interface table of contents Toronto, Ontario,Canada Pages: 255 - 260, ISBN:0-89791-213-6
[Vilimeck, Zimmer, 2007] Vilimek, R., Zimmer, A. (2007) Development andEvaluation of a Multimodal Touchpad for Advanced In-VehicleSystems, Engin. Psychol. and Cog. Ergonomics, HCII 2007,Springer-Verlag Berlin Heidelberg
[Volvo Strategic Planning, 2008] Fast P. & du Val D. (2008) Volvo Cars Cor-poration Strategic Planning, Gothenburg
[BBC, 2008] Jeremy Clarkson, BBC TopGear, Season 1, episode 5, Originalairdate November 17, 2002http://www.youtube.com/watch?v=2p8BtTUwP44
90
Appendix A
Interview questions
Hi!I work on a master thesis project in interaction design at Chalmers (for
Volvo HMI, Volvo Design and Interaktionsbyran) that aims to evaluate the“Touchpad as interaction input control for use if in-vehicle infotainment system”.I.e. research if it might be suitable to use a touchpad to control the infotainmentand/or more systems in the car.
I would like to conduct a short informal interview with you to get a bet-ter/wider understanding of the problem from you angle, since people at differentdepartments generally seems to have diverging opinions, naturally because theywork on different premises. What I am interested in are some answers to thesequestions from your point of view.
Here are the questions:
• What have you discussed so far on this issue?
• What could be the motivation to make this change?
• What do you think of the concept?
• What are your expectations?
• What are your concerns?
• Do you know any other person that I should contact about this?
91
APPENDIX A. INTERVIEW QUESTIONS
92
Appendix B
Questionnaire
The questionnaire that was used to evaluate the concept in the third interation isattached on the following pages. It has been translated to english from swedish.
93
Questionnaire: touchpad
June 9, 2009
1 Background1. Gender
Woman Man
1. How old are you ?20-30 31-40 41-50 51-60 61-70
Comment:
2. Do you consider yourself interested in technology?Not at all Very much Don’t know
Comment:
3. What is your education and occupation?
4. Do you have a car?Yes No Don’t know
Comment:
5. Do you want to buy a car or switch car within the next 5 years?Not at all Very much Don’t know
Comment:
6. Have you used the infotainment system in a car 1 ?Not at all Very much Don’t know
1Infotainment is a fusion of the words information and entertainment. It is used an en-compassing term to describe multimedia- and navigationssystems, cd, dvd etc. in cars .
1
Comment:
Do not answer the following question if you answered “Not at all” above.
(a) Was it a positive experience?Not at all Very much Don’t know
Comment:
7. Do you think the infotainmentsystem will have a major influence on youchoice of car?Not at all Very much Don’t know
Comment:
2 Touchpad1. Do you like using/working with a touchpad (e.g. laptop, ipod touch etc.)?
Not at all Very much Don’t know
Comment:
2. Have you used a touchpad with multitouch2 function?Not at all Very much Don’t know
Comment:
Do not answer the following question if you answered “Not at all” above.
(a) Do you like using/working with a multi-touch touchpad?Not at all Very much Don’t know
Comment:
3 Questions for discussion1. Do you like the touchpad (form, function)?
2. What do you think about the home-button (form, function)?
3. What do you think about the placement of the?2The touchpad recognizes if more than one finger has been placed in it, as opposed to
regular touchpads that only recognizes one point of touch
2
4. How do you appreciate the different interactions?
(a) panning e.g. map and internet
(b) gestures e.g. back or forward on a web page
(c) character input
(d) zoom e.g faces and map
5. Compare this system to others you have used; what do you think aboutthis system?
6. Do you think it can be used by both novice or experts?
7. Do you think it can be used by drivers with differnt lengths?
8. Do you think this can be used by both right- and left-handed?
9. Would you like to have this as the only input device in your car?Not at all Very much Don’t know
Comment:
10. Would you like to have this as a compliment to buttons and other con-trols?Not at all Very much Don’t know
Comment:
4 What do you think about the potential of theconcept
4.1 Home button1. What do you think about the function of the home button?
Not at all Very much Don’t know
Comment:
3
4.2 Marking and selection
Figure 1: Marking and selection
1. Do you think the way to select an element in the display works well?Not at all Very much Don’t know
Comment:
4.3 Zoom
Figure 2: Zoom in the map
1. Do you think the way zoom works in the map is good?Not at all Very much Don’t know
4
Comment:
2. Do you think zoom for searching information (e.g. finding music or con-tacts) is good?Not at all Very much Don’t know
Comment:
4.4 Panorering
Figure 3: Pan
1. Do you think panning for searching information (e.g. finding music orcontacts) is good?Not at all Very much Don’t know
Comment:
2. Do you think panning for searching information in the web browser (e.g.navigating web pages or searching) is good, as it works in the concept?Not at all Very much Don’t know
Comment:
5
4.5 Gestures
Figure 4: Gestures
1. Do you like the way gesture works for interacting with simple dialog boxes?Not at all Very much Don’t know
Comment:
4.6 Input1. Do you like the way character input is done?
Not at all Very much Don’t know
Comment:
6
