Using a Mahl-Stick as a 2-Dimensional Spatial … · entertainment, engineering, education,...

81
School of Information Technology and Mathematical Sciences Division of Information Technology, Engineering and the Environment Research Proposal for Bachelor of Computer Science (Honours) Using a Mahl-Stick as a 2-Dimensional Spatial Augmented Reality Input Device Matthew McDonald [email protected] Supervisors: Bruce H. Thomas, Ross T. Smith Date of submission: 5 / 2 / 2015

Transcript of Using a Mahl-Stick as a 2-Dimensional Spatial … · entertainment, engineering, education,...

School of Information Technology and Mathematical Sciences

Division of Information Technology, Engineering and the Environment

Research Proposal for Bachelor of Computer Science (Honours)

Using a Mahl-Stick as a 2-Dimensional Spatial

Augmented Reality Input Device

Matthew McDonald [email protected]

Supervisors: Bruce H. Thomas, Ross T. Smith

Date of submission: 5 / 2 / 2015

Copyright © 2015

Matthew McDonald

All rights reserved.

i

ABSTRACT

This dissertation presents an exploration of the effect that a mahl-stick, a traditional tool used

to support the brush hand in painting and signwriting, has in simple applications in a Spatial

Augmented Reality context. Spatial Augmented Reality uses digital projectors to add

computer generated images to real-world objects interactively and at run-time, and is used in

entertainment, engineering, education, industrial design, business collaboration and

manufacturing. Input devices to interact with these applications need to be developed and

improved. To this end pointing (selection) and steering (drawing) tasks are examined to see

how a mahl-stick can be used in a SAR application.

To evaluate this, two user studies were conducted in which participants were asked to

perform pointing and steering tasks using a stylus, with and without the aid of a mahl-stick.

Participants were measured on time, accuracy, and the number of errors made whilst

performing these tasks. Participants rated their opinions on their performance in terms of

ease, accuracy and speed of performing the tasks. Participants technique and fatigue were

also monitored.

This dissertation has focused on simple, small-scale and straight line pointing and steering

tasks on a vertical surface. Mahl-sticks have demonstrated an impaired performance for

pointing tasks, however a preference for the use of a mahl-stick for steering tasks in such

conditions was revealed. Fatigue was found to have an influence on task performance and

user preferences and it took little time before it affected users in a negative fashion. An

artefact of this research is a new AR input device.

ii

DECLARATION

I declare that:

this thesis presents work carried out by myself and does not incorporate without

acknowledgment any material previously submitted for a degree or diploma in any

university;

to the best of my knowledge it does not contain any materials previously published or

written by another person except where due reference is made in the text; and all

substantive contributions by others to the work presented, including jointly authored

publications, is clearly acknowledged.

Matthew McDonald

5th

February 2015

iii

ACKNOWLEDGMENTS

I would like to thank my supervisors, Bruce Thomas and Ross Smith, for their advice,

opinions and support over the past year. Thanks to everybody in the Wearable Computer Lab:

Michael Marner, James Baumeister, James Walsh, Andrew Irlitti, Neven Elsayad, and Tim

Simon - you're all awesome. I would also like to thank my employers and colleagues for

providing me the time I have needed to complete my studies. Finally, I would like to thank

my family and friends for their support and good humour.

iv

TABLE OF CONTENTS

1. INTRODUCTION................................................................................................................ 1

1.1. The Problem .................................................................................................................... 2

1.2. Research Question ........................................................................................................... 2

1.3. Contributions ................................................................................................................... 3

1.4. Dissertation Structure ...................................................................................................... 3

2. BACKGROUND .................................................................................................................. 5

2.1. Augmented Reality .......................................................................................................... 6

2.2. Spatial Augmented Reality .............................................................................................. 8

2.3. Tracking ........................................................................................................................ 13

2.4. Spatial Augmented Reality User Interfaces and Input .................................................. 16

2.5. Fitts' Law ....................................................................................................................... 20

2.6. Steering Law ................................................................................................................. 23

2.7. Mahl-Sticks ................................................................................................................... 26

2.8. Summary ....................................................................................................................... 28

3. RESEARCH METHOD .................................................................................................... 29

3.1. Pointing Task Study Methodology ................................................................................ 30

3.1.1. Goal ........................................................................................................................ 30

3.1.2. Hypothesis .............................................................................................................. 31

3.1.3. Pointing Task Study Design ................................................................................... 31

3.2. Steering Task Study Methodology ................................................................................ 36

3.2.1. Goal ........................................................................................................................ 36

3.2.2. Hypothesis .............................................................................................................. 36

3.2.3. Steering Task Study Design ................................................................................... 37

3.3. User Study Environment ............................................................................................... 40

3.3.1. Projection System ................................................................................................... 40

3.3.2. Tracking System ..................................................................................................... 43

3.3.3. Stylus ...................................................................................................................... 45

3.3.4. Mahl-Stick .............................................................................................................. 46

v

4. ANALYSIS ......................................................................................................................... 47

4.1. Pointing Task User Study .............................................................................................. 49

4.2. Steering Task User Study .............................................................................................. 52

4.3. Qualitative Results ........................................................................................................ 55

4.3.1. Observations of Participants ................................................................................... 55

4.3.2. Questionnaire Results ............................................................................................. 57

4.4. Summary of Results ...................................................................................................... 59

5. CONCLUSION .................................................................................................................. 60

5.1. Pointing Tasks ............................................................................................................... 60

5.2. Steering Tasks ............................................................................................................... 61

5.3. Future Directions & Final Comments ........................................................................... 61

REFERENCES ....................................................................................................................... 63

vi

LIST OF ABBREVIATIONS

2D 2-Dimensional

3D 3-Dimensional

AR Augmented Reality

CAVE Computer Automated Virtual Environment

CRT Cathode Ray Tube

HMD Head-Mounted Display

GUI Graphical User Interface

IR Infra-Red

LED Light Emitting Diode

RFID Radio-Frequency Identification

SAR Spatial Augmented Reality

SID Spatially Immersive Display

TUI Tangible User Interface

VR Virtual Reality

vii

LIST OF FIGURES

Page Figure

6 1.1 The Mixed Reality Continuum

26 2.1 A Mahl-stick in Use

32 3.1 Arrangement and Size of the Projected Targets in the Pointing Task Study

32 3.2 The Preferred Height of the Targets in Relation to a Participants' Height

33 3.3 Performing the Pointing Task With and Without the Mahl-Stick

34 3.4 Example Pointing Task

37 3.5 Arrangement and Size of the Projected Targets in the Steering Task Study

38 3.6 Example Steering Task

39 3.7 Performing the Steering Task With and Without the Mahl-Stick

40 3.8 Occlusion from a Single Projector

41 3.9 Description of Virtual Rear Projection

42 3.10 Showing Virtual Rear Projection in Action

44 3.11 The Placement of Motion Capture Cameras Around the Scene

44 3.12 The Combined Coordinate System for the Cameras and Projectors

45 3.13 The Stylus used in the User Studies

46 3.14 The Mahl-Stick used in the User Studies

viii

LIST OF FORMULAE

Page Formula

20 1 Fitts' Law

20 2 Fitts' Law as derived originally by Fitts (1954)

20 3 Welford Formulation of Fitts' Law

20 4 Shannon Formulation of Fitts' Law

21 5 MacKenzie & Buxton's First Formulation of Fitts' Law (1992)

21 6 MacKenzie & Buxton's Second Formulation of Fitts' Law (1992)

21 7 Kopper et al.'s Formulation of Fitts' Law (2010)

21 8 Appert et al.'s Formulation of Fitts' Law (2008)

21 9 Yang and Xu's Formulation of Fitts' Law (2010)

22 10 Zhang et al.'s Formulation of Fitts' Law (2012)

23 11 Steering Law as derived originally by Accot & Zhai (2001)

23 12 Accot & Zhai's Formulation of Steering Law (2003)

23 13 Grossman & Balakrishnan's Formulation of Steering Law (2005)

23 14 Pastel's Steering Law for steering around corners (2006)

24 15 Zhou & Ren's Formulation of Steering Law (2010)

24 16 MacKenzie et al.'s Throughput Metric (2001)

1

1. INTRODUCTION

Augmented Reality (AR) is the integration of computer-generated sensory information

directly into the real world that occurs in real-time, is interactive and aligns real and virtual

objects (Azuma 1997). This differs from Virtual Reality (VR) in that AR uses the real

environment whereas VR creates an entirely virtual environment (Milgram & Kishino 1994).

Spatial Augmented Reality (SAR) is a form of AR in which computer-generated imagery is

projected directly onto objects in the real world, most often achieved through the use of

projectors, flat panel displays and smart boards (Raskar et al. 1998). SAR has applications in

a large range of fields including entertainment, education, industrial design, business

collaboration, ordinary workflow improvements, and manufacturing.

Interactions with a SAR system can be achieved by various means but is better achieved with

physical tools, props and by registering the user's body movements (Mine et al. 1997, Marner

et al. 2011). With many different applications, there are a wide variety of tools and methods

to interact with these systems. One such simple tool currently used is the stylus. The stylus

can be used as a pen and to provide more precision when performing selection tasks. These

two actions can be described as steering tasks (dragging the stylus along a specific path

across the surface of an object) and pointing tasks (clicking on a specific point on the surface

of an object) respectively.

One common steering task is drawing a line. Drawing lines is an integral part of the creative

process as it is part of seeing and understanding the subject matter itself. Combining lines is

an effective way to describe and share what is contained within a person's imagination, and

these combinations often take the forms of pictures and text. A straight line can be defined as

a path between 2 endpoints. This thesis is an examination in the enhancement of free-hand

drawing through the use of a mahl-stick, a supporting rod usually 1 meter in length for the

brush hand used traditionally in painting and signwriting, within the context of Spatial

Augmented Reality.

2

1.1. The Problem

One problem that can emerge in undertaking pointing and steering tasks lies in the accuracy

obtained when performing them. It can be difficult to steer lines along a precise path by hand

or to select a specific point on the surface on an object. The software in drawing systems can

interpolate between a drawn line and what the user intended to draw by simplifying, adjusting

and moving the drawn line to be neater. However it is impossible to know with certainty

which features of a drawn path are desired and which features are not. Alternatively a

person's skill in drawing can improve to the point where they will draw lines exactly how

they want; however obtaining this level of skill can take years of practice.

A mahl-stick reduces the difficulty in drawing accurate lines by providing support for the

hand performing the drawing task, and by decreasing the amount of movement the arm needs

to make to draw a line. A mahl-stick is held by the non-drawing hand roughly horizontal,

with the far end resting upon a stable surface. This provides a stable platform for the brush

hand to rest upon. Normally when drawing the shoulder, elbow, wrist and fingers all need to

move. If the drawing hand rests on a mahl-stick only the wrist and fingers need to move,

decreasing the number of body parts that need to move and focusing attention toward the

most precise: the fingers. My knowledge of mahl-sticks was obtained from my training in

Signwriting at the Gilles Plains TAFE, South Australia.

1.2. Research Question

The focus of this research is to evaluate the effect that a mahl-stick has when performing

simple pointing and steering tasks in Spatial Augmented Reality applications when using a

stylus.

There are two sub-questions in relation to this research question:

How do novice and experienced users compare with the use of a mahl-stick in

performing pointing and steering tasks.

What learning effect is observable in using a mahl-stick.

An artefact of this research is a new handheld SAR drawing device.

3

1.3. Contributions

There has been a trend in research to improve the accuracy in pointing and steering tasks

algorithmically or with more complex and sophisticated tracking and input devices. This

research is focused on a proven, simple, centuries-old technique to improve accuracy by

providing physical support to the user's drawing hand instead by way of a mahl-stick.

The sub-questions of the research also address the learning effect of pointing and steering

tasks. The learning effect is usually excluded from results in the analysis following studies or

pointed out as a mitigating variable in the results; with this research I aimed to specifically

look into the effect learning has in completing tasks.

Furthermore this research provides an analysis of mahl-sticks in completing pointing and

steering tasks. To the best of my knowledge this is something that has never been

investigated before. The method chosen to achieve this is with analysis of captured time and

movement data, user opinions, and observation of participants in a user study, with the

difficulty of the tasks tested against Fitts' Law and the Accot-Zhai Steering Law.

Lastly, an artefact of this research is a new input device for SAR applications.

1.4. Dissertation Structure

This thesis is structured as follows. In Chapter 2 the body of knowledge upon which this

research is built is explored. A definition of Augmented Reality (AR) is given and categories

of AR systems are described. Following that a definition of Spatial Augmented Reality

(SAR) is provided, several successful SAR implementations are described, and methods and

pitfalls in their development are explained. Various methods for tracking and virtual-real

object alignment in AR systems are listed next, followed by a description of various SAR

user interfaces and input devices. Fitts' Law and the Accot-Zhai Steering Law, two formulae

used to compare input devices on their capacity to perform pointing and steering tasks, are

detailed. Lastly mahl-sticks are described, and Chapter 2 is concluded with a summary.

To answer the research questions, two concurrent user studies were conducted. The

methodology for both user studies is described in Chapter 3. First described is a pointing task

study was conducted to test the performance of pointing tasks with and without the aid of a

4

mahl-stick. This is followed by the methodology for a steering task study, again conducted to

compare performance with and without the stick. Both user studies were ran simultaneously,

using the same equipment and environment. The chapter is concluded with detail describing

the projector system and its' calibration, the tracking system in use, and the construction and

use of the stylus and mahl-stick used in the studies.

In Chapter 4 the analyses of the results from both user studies is described. Both quantitative

and qualitative data were recorded during the user studies. The qualitative data for the

pointing task study is analysed first followed by that of the steering task study. The

quantitative data captured during the study was obtained by observation of the participants,

and by a questionnaire following the conclusion of the studies. The observation data is

analysed before the results of the questionnaire. Chapter 4 is concluded with a summary of

the results.

The thesis is concluded in Chapter 5. The body of knowledge provided in Chapter 2 is

summarised in brief. The methodology and analysis of results of the pointing task study are

described, followed by that of the steering task study. The chapter is concluded with future

directions for research of mahl-sticks in SAR systems, and final comments regarding this

dissertation.

5

2. BACKGROUND

This section of the thesis serves to define the context of the research described in this

dissertation by providing a review of relevant literature. Firstly Augmented Reality and

Spatial Augmented Reality are defined, and various implementations of these systems is

provided. To this end, various pitfalls and calibration techniques are described. This is

followed by a description of various tracking techniques. Whilst tracking within SAR

systems was not a focus in this research, it is a fundamental aspect of interactive SAR

systems such as those in which a stylus is used. This is followed by an exploration of various

user interfaces and input devices in SAR systems.

Fitts' Law and the Steering Law are then reviewed to provide quantitative measurements for

comparing the difficulty of user input systems. These are mathematical formulae that are

standard in comparing non-keyboard input devices but possess a wide variety of formulations

and limitations. As such these were examined as they were used to help analyse the results

obtained in this research.

This is followed with a description of mahl-sticks in greater detail to explain the manner in

which they are used. This chapter is concluded with a summary of the information described

in the above.

6

2.1. Augmented Reality

In 1965 Ivan Sutherland wrote in The Ultimate Display of a computer device that is able to

control the existence of all matter in a room, creating virtual objects indistinguishable from

real world objects. Of this he wrote:

The ultimate display would, of course, be a room within which the computer can

control the existence of matter. A chair displayed in such a room would be good

enough to sit in. Handcuffs displayed in such a room would be confining, and a bullet

displayed in such a room would be fatal.

This described all physical reality being altered and controlled by digital information.

However he also described a system in which the real sensory information of objects is

altered digitally. This integration of computer-generated sensory information directly into the

real world in real-time is known as Augmented Reality (AR) (Azuma 1997). Azuma et al. in

2001 provided a definition of AR that it:

combines real and virtual objects in a real environment;

runs interactively, and in real time; and

registers (aligns) real and virtual objects with each other.

The Mixed Reality Continuum (Figure 1.1) described by Milgram and Kishino (1994) defines

AR in relation to the real environment and Virtual Reality (VR). At one end lies the real

environment, and at the other end lies Virtual Reality which is entirely computer generated.

The line between them represents the gradual inclusion of computer generated information

and the resulting exclusion of the real environment. AR is placed towards the real

environment, indicating that the focus is on adding digital information to the real world.

Figure 1.1: The Mixed Reality Continuum (Milgram & Kishino 1994)

These augmentations can be for any sense: sight, hearing, taste, smell, etc., though the most

prevalent augmentations are sight-based. AR visualisation can be achieved with head-

7

mounted displays (HMDs), projectors, or specialised display surfaces such as tablets and

screens.

HMDs come in two flavours: Optical See-Through where augmentations are projected onto

glasses or directly on the retina of the eye by lasers, and Video See-Through where

augmentations are added to video displays. HMDs can also be monoscopic where both eyes

receive the same image, or stereoscopic where each eye receives an adjusted image to allow

for the illusion of depth. The first AR system was created by Sutherland in 1968 and named

'The Sword of Damocles', and it also featured the first HMD (Sutherland 1968). The HMD

was stereoscopic optical see-through and suspended from the ceiling, tracking and updating

wireframe objects as the user moved.

Whilst it is easy to add light and translucent objects to an optical see-through scene, adding

shadows and solid objects to the environment can be much more challenging. Bimber and

Fröhlich (2002) developed a technique where the light source in the environment is replaced

by projectors. Objects can be made to appear solid by not projecting light where the virtual

object would go for the viewer's perspective. Shadows can also be created by projecting only

some light onto the environment where the shadow would be.

Completely virtual objects displayed through HMDs provide no haptic feedback to users. The

X'tal Vision system (Inami et al. 2000) uses a plane of retro-reflective glass as an optical see-

through HMD and a head-mounted projector to help create viewer-perspective augmented

objects that can be touched and interacted with in real time, using front projection techniques.

8

2.2. Spatial Augmented Reality

Using projectors, flat panel displays and smart boards to project directly onto real objects in

the real world is known as 'Spatial Augmented Reality' (SAR) (Raskar et al. 1998). The key

advantage of SAR systems is that the user need not encumber themselves with potentially

expensive or heavy HMDs because the augmentations are provided directly into the

environment itself. As such much of the research and development of AR systems has moved

towards SAR (Lantz 1996). SAR currently has a wide variety of applications across many

fields such as in education (Bimber & Raskar 2005), entertainment (Oda & Feiner 2009),

business collaboration and workflow (Wellner 1993), industrial design (Marner et al. 2011),

and review (Verlinden et al. 2009).

The seminal work on the CAVE (Computer Automated Virtual Environment; Cruz-Neira et

al. 1993) made use of projection technology on the surface of the walls of a room to act as

shared workspaces in a collaborative workplace environment. The users of the system wore

tracked HMDs that provided further augmentations tailored to each individual's requirements.

User studies of this system revealed an interesting behaviour in regards to how people dealt

with occlusion: people quickly resolved occlusions by removing the offending object or body

part from view. Occlusions caused by other people or objects outside of their ability to move

were better controlled by a single user controlling the proceedings such that occlusions were

unlikely to occur.

Concurrently developed, the DigitalDesk was a system that aimed to combine the physical

work desk with a virtual one to improve workflow (Wellner 1993). The CAVE and

DigitalDesk systems led to further research on using AR to improve collaboration and

business workflow and by 1996 a panel at SIGGRAPH '96 organised by Ed Lantz had

already noted a trend away from HMDs towards SAR, then termed Spatially Immersive

Displays (SID). The Office of the Future (Raskar et al. 1998) described a vision for future

workspaces where every surface is a possible interactive surface. The research also

introduced imperceptible light patterns to calibrate camera-projector pairs and used blending

techniques to combine the overlapping imagery from different projectors.

Another early SAR system was the Luminous Room (Underkoffler et al. 1999) which made

use of every surface of a room as a projector display surface and any object within that room

could be given a passive role to perform virtual actions.

9

Perhaps the first system that allowed people to peer around corners and take advantage of

their peripheral vision to increase the immersion of an AR system was Being There (Low et

al. 2001). They were able to create an environment out of white Styrofoam blocks and use

projectors to change the scene to another environment and also allowed users to walk through

the environment in real time.

The Shader Lamps system (Raskar et al. 2001) could change the complete appearance of

complex objects regardless of shadows and self-occlusion. However the system was

dependent upon the objects possessing an amenable surface and required dark ambient light

to achieve its full effect. Dynamic Shader Lamps (Bandyopadhyay et al. 2001) allowed the

user to alter the appearance of objects in real time. A stylus with a magnetic tracker and 3

LEDs could be tracked within a predefined computational bounding box corresponding to

real space although the system suffered from latency issues.

Raskar and Low (2001) examined the implementations of three successful early SAR

systems: Shader Lamps, Tracked Object Illumination, and Being There. From these they

were able to discern some common benefits and limitations that existed between the three

systems:

All did not constrain the user is to one position or perspective.

All forewent HMDs, and only made use of head tracking if they wanted to

dynamically alter the appearance of objects given the viewer's location.

All were dependent upon the surfaces being projected upon to having properties

amenable to projections.

All had to carefully position the projectors to minimise shadow occlusion.

These observations have tended to remain true for the development of SAR systems since.

Users of SAR systems are more concerned with the brightness, contrast and saturation of the

projection images than they are with any usability concerns (Brooks Jr. 1999). Laser

projectors can provide superior resolutions and colour saturation than other kinds of

projectors, however they are far more expensive as well (Schwerdtfeger et al. 2008). Laser

projectors are also unable to do much towards increasing the contrast of the projected images.

Less expensive CRT and LED projectors can overlap their projections onto surfaces, and in

that manner combine into a single brighter image (Majumder & Welch 2001; Bimber &

10

Emmerling 2006). A side effect of combining projections is the final image can have a higher

focus, improving the image quality. Projections from multiple projectors had been combined

since at least the Office of the Future (Raskar et al. 1998), however this was done in order to

create larger images rather than brighter ones. The technique described by Raskar et al.

(1999) used projected bands from each projector being decoded individually to discover the

overlaps. The projections were then alpha blended and stitched together to try and create

seamless images.

Shadow occlusion can be reduced or even eliminated by using several redundant projections

focused upon the same plane. If part of the image from one projector is occluded, another

projector can still project onto where the image was (Summet et al. 2005). The occluding

object can be detected and be not projected onto in real time.

Aligning projections with real world objects has proven a difficult task. Early systems

required a time-consuming manual calibration. The Office of the Future (Raskar et al. 1998)

introduced imperceptible light patterns to calibrate camera-projector pairs, making the

process slightly easier and faster. Zhang (2000) developed a technique to calibrate a

projector-camera pair by calculating the intrinsic and extrinsic parameters of either the

camera or the display surface and then moving either one or the other a minimum of 2 times.

A self-correcting projector-camera pair technique was developed by Raskar & Beardsley

(2001) which was able to perspectively-correct keystone the projection to the planar target

surface.

The iLamps system (Raskar et al. 2003) improved calibration and registration on multiple

projectors by using colour banding and did not require projectors and cameras to be aligned

on the same axis. Light sensors embedded into the object being projected upon can

automatically calibrate the projections with the surface without the use of a camera (Lee et al.

2004).

Another challenge found with SAR is that projecting the desired appearance straight onto an

arbitrary object often gives unsatisfactory results as the original colouration of objects with

non-Lambertian surfaces is visible through the projection. Grossberg et al. (2004) developed

a technique for capturing the colour channels of the object and then applying a compensation

filter onto the projection image that resulted in the objects own appearance being washed out

with the application of new light. Structured light patterns have been used to perform

11

radiometric compensation in projector-camera systems to help improve the colour quality

(Zollman & Bimber 2007). Such light structures can be cycled faster than the human eye can

perceive.

It has proven difficult to provide view-dependent projections for different users in multi-user

SAR systems. The regular use of projections provides for the same image to been seen by all

viewers from all angles. The Being There system (Low et al. 2001) which allowed viewers to

walk through the SAR environment provided HMDs to create the appearance of perspectively

correct windows through solid objects. Another method was developed by Agrawala et al.

(1997) in which shutter glasses provided low bandwidth images to each eye for up to two

people.

Images have also been projected onto inverted mirrored cones whilst the viewers wear HMDs

tracking their view orientation, allowing the users to see perspectively correct computer

generated imagery in a single location (Bimber et al. 2003). This has been used to allow

people to view fossil specimens, and see the organisms’ various tissues layered upon it

(Bimber et al. 2002). A similar idea has been to project directly onto artwork in galleries,

allowing viewers to see one at a time the various drafts and original forms of the paintings as

the artist worked on them before settling on the final appearance (Bimber et al. 2005).

Getting projections on all sides of all desired objects can be a challenge. Whilst placing more

projectors throughout the environment is an option, it can be expensive and infeasible with

certain environments and applications. In 2001 Pinhanez developed a technique in which a

movable mirror is used to project onto surfaces outside the line of sight of a projector,

without having to recalibrate the system. Having users hold the projectors is another method

to augment real world objects in their physical location (Beardsley et al. 2005), and such

projectors can also be used as an input device.

SAR can greatly decrease the cost and time taken to prototype in industrial design. Projected

light is faster and cheaper to change than clay modelling and 3D printing, and ideas can be

trialled ad hoc efficiently without having to alter the actual surface of the object and the

environment that contains it. Being able to cycle through different appearances and

configurations also reduces the number of physical objects needing to be created and reduces

storage space required for prototypes (Marner et al. 2011).

12

Improving the quality of SAR systems has been shown to improve the sense of object-

presence which can improve the experience that people have interacting with the system

(Stevens et al. 2002). This improvement has been demonstrated to help improve the learning

and performance of tasks (Witmer & Singer 1998). However the sense of touch can reduce

object-presence when the viewer becomes aware that the augmentations do not possess the

haptic properties the visual augmentations suggest (Bennett & Stevens 2005).

13

2.3. Tracking

Most interactive SAR systems need to track the position and orientation of objects in order

for the real and virtual objects to properly align (Marner 2013). If props and simple tools are

being used to interact with the SAR system, these need to be tracked too. Similarly, tracking

users can allow for view-dependent projections. This section provides a brief overview of

tracking techniques in SAR systems.

There are six main methods to track objects in the real world: magnetic, acoustic, inertial,

mechanical, optical, and radio / microwave sensing (Bhatnagar 1993, Welch & Foxlin 2002).

Magnetic trackers measure a local magnetic field vector using magnetometers or

electromagnetic coils. These trackers are usually lightweight, avoid line-of-sight

issues and possess high update rates, however they are all vulnerable to distortion

from environmental magnetic, electromagnetic, and metallic or ferromagnetic fields.

Acoustic trackers measure the time it takes ultrasonic pulses to reach receivers to

track objects over time. These trackers are lightweight but are limited by the low

speed of sound, suffer from echoes, require line-of-sight and are subject to disruptions

in air temperature and humidity.

Mechanical trackers measure the position and orientation of objects attached to the

end of a movable mechanical arm. These trackers are simple to construct (the Sword

of Damocles system built by Sutherland in 1968 used mechanical tracking) but the

limitation that objects be in range of the arm is so severe that they are largely obsolete

in modern SAR systems.

Inertial trackers make use of gyroscopes and accelerometers to capture the movement

in 3 linear axes. From these a transformation matrix is calculated to determine the

position of the object in the real world. These trackers are self-contained, have high

update rates and face no interference from electronic fields and ambient noise.

However inertial trackers suffer from jitter and drift, in which even a small bias error

in one axis will cause the estimates to drift vast distances in only a short time.

Optical trackers track visual cues on objects or in the environment. They possess high

update rates, large working volumes and are unaffected by the presence of metals and

electromagnetic fields in the environment, however they also suffer from line-of-sight

issues and can be affected by ambient light and infrared radiation.

14

Radio and microwave tracking embeds small tags that emit radio or microwave

electromagnetic waves which can then be tracked. This offers a greater range than

magnetic trackers and are largely unaffected by wind and air temperature, however

the waves are rapidly attenuated in water: essentially this makes the human body

opaque for this tracking technique.

All tracking techniques share a problem in that latency exists between the time a person

makes an input and when the system registers it. Alleviating this by predicting future

movement is possible only to a limited extent as humans are unpredictable. For this reason a

tracking technique which has a high update rate is usually desired. Optical tracking is the

most common method used today in part because of its high update rate but also because they

are relatively simple to implement and SAR systems already tend to involve cameras.

Fiducial markers, simple unique high-contrast patterns, are a popular optical tracking

technique. Many AR systems detect fiducials in the environment and display projections

relative to their positions, and users are intuitively able to manipulate them to perform tasks

(Kato & Billinghurst, 1999). Software libraries such as ARToolKit are able to detect fiducials

in the environment to determine position, and the ARToolKitPlus is an updated version able

to run on mobile devices (Wagner 2007).

It has been has been demonstrated that placing fiducial markers within the environment and

the camera / tracker placed on the object is more accurate than the reverse as the distances

between the markers is greater allowing for greater precision (Bhatnagar 1993). However for

most SAR systems this is impractical.

Another form of fiducial marker is Random Dot Markers, unique patterns of dots that can

easily be rotated and arranged into any shape (Uchiyama & Saito 2011). One advantage these

markers have over other types is that they don’t require hard edges, and they still work even

when partially occluded or incomplete. They can also be deformed, and the deformation can

be detected and measured (Uchiyama & Marchand 2011).

The Bokeh effect has also been used to decrease the size of fiducial markers detected by

ordinary cameras down to 3mm from a distance of up to 4m away by taking advantage of the

effect that occurs when out of focus scene point is focused into a blur on the camera sensor

(Mohan et al. 2009).

15

Fiducial markers themselves are obtrusive, often don't work if partly occluded, and require

high ambient light to be effective whilst SAR systems perform best in low light conditions

(Marner et al. 2011). As such invisible or active (reacting to the immediate environment)

markers are usually preferred in SAR systems.

There are several different techniques used to create invisible fiducials. Grundhöfer et al.

(2007) used the green channel, to which the human eye is most sensitive, to encode fiducials

into a projected image one frame, and its inverse the next. As this process repeats, the human

eye is unable to detect the fiducials whereas a computer can. However this technique assumes

the presence of the green channel throughout the entire image to place the fiducial in the first

place. Using other channels can result in a more noticeable decrease in brightness.

Infrared (IR) fiducials are another way to create invisible fiducials though care needs to be

taken so that other sources of infrared light, such as the sun or fluorescent lighting, do not

reduce their clarity (Nakazato et al. 2005). Infrared can be observed by itself or with visible

colours as well. Park & Park (2004) used a colour camera and an IR camera focused on

precisely the same location through the use of a half-silvered mirror to detect infrared

fiducials whilst obtaining the visible spectrum image as well. Infrared fiducials have been

placed on the ceilings of interiors to navigate users through interior spaces (Nakazato et al.

2008).

Colour blobbing is another optical tracking technique in SAR systems which allows for the

tracking of objects of any colour in real time (Rasmussen et al. 1996). Colour blobbing is a

simple image processing technique in which the centre of unique blobs of colour is

discovered. Multiple objects can be tracked over time by performing this across frames.

Embedding optical and radio tags into objects is another tracking technique gaining

popularity in AR applications. In the Prakash system (Raskar et al. 2007) IR tags emit coded

signals that photosensors are able to detect and from there calculate the motion, orientation

and illumination of the tagged points. They used this system to embed virtual objects into a

live capture of a scene. RFID tags have been embedded into objects for self-description; once

detected the system sends a grey code to the RFID which decrypts it and sends it back. From

this the system is able to determine the location of an object and project onto it (Raskar et al.

2004).

16

2.4. Spatial Augmented Reality User Interfaces and Input

As previously stated, SAR has many applications in a wide range of fields including

education, entertainment, business collaboration and industrial design. As AR augments

objects in the real world, there is no one size fits all approach for interacting with these

systems and the right tool has to be selected for the task at hand (Brooks Jr. 1999).

Interactions within SAR systems is better done with physical tools and manipulating props, or

by capturing the movement of the user's own body (Marner 2013). This section describes

interfaces and input methods to interact with SAR systems.

The cognitive load of using a tool or system is increased with the number of different

functions given to it. Likewise the cognitive load of a system is increased the more the

system is capable of performing (Marner & Thomas 2010). Using an all purpose tool is more

difficult for users than using their own hands and body movement to interact with a fully

virtualised system (Mine et al. 1997). Similarly it is easier for people to interact with a virtual

object when they possess a physical representation or model of that object in their hands, with

users preferring simple props to complex ones. For example, in a prop used to navigate a 3D

image of a human brain, neurosurgeons preferred to use a simple ball rather than dolls head

(Hinckley et al. 1994).

2-Dimensional GUIs can be augmented into the real world in a multitude of ways (Feiner et

al. 1993). GUIs are already understood by users due to their prevalence in desktop and

mobile computing and there exist numerous variations on GUI interactions (Bier et al. 1993).

However the traditional techniques used for navigating 2D spaces, such as panning and

tilting, are often disorientating for users in AR applications. Physical navigation (people

moving their eyes, hands, body, the object, the environment, etc.) is preferred by users if their

movement is not restricted by corded or immobile input devices (Ball, North & Bowman

2007).

Many GUI manipulation tasks are pointing or steering tasks in nature: 'clicking' on objects or

moving along certain paths. Both pointing and steering tasks can be virtualised through

another surface or device, performed directly, or performed with a distal device such as a

laser pen. There are other methods to interact with GUIs in AR that do not necessary include

point and steering. In the Tinmith System (Piekarski & Thomas 2002), the user wore

electronic gloves in which finger and hand movements could manipulate GUI components.

17

SixthSense (Mistry & Maes 2009), a neck-worn pendant projector-camera system, captures

hand gestures and interactions made in front of the user. Shadow Reaching (Shoemaker et al.

2007) is a form of distal pointing which makes use of the user's shadow to interact with

distant virtual objects.

Motion Swarms (Nguyen et al. 2006) is a system in which an audience acts as input by

creating a virtual swarm of particles controlled by movement in the audience. They used this

in applications in which an audience controlled a virtual beach ball, played music, and

painted a picture.

Cao & Balakrishnan (2003) devised a system in which the movement of a handheld projector

affords certain kinds of input, whilst a pen provided for alternative forms of input.

Laser pointers are another type of input device used in AR applications. An early study of

laser pointers revealed they are slower than using a traditional keyboard and mouse, and

suffer from jitter (Olsen Jr. & Nielsen 2001). One way to decrease jitter and improve

accuracy is to change the way in which a user holds the pointer. An 'arrow-holding' technique

in which the user positions the laser pointer out in front of their eyes such that they look

down the length of the pointer at the target has been demonstrated to offer the overall best

results for accuracy across large surfaces at the cost of the user's arm fatigue (Jota et al.

2010). Another method to decrease jitter and improve accuracy is to create a virtual bubble

around the point in which that if the point remains within it, the position within the bubble is

stable (Forlines et al. 2005).

Tracking laser pointers in the real world can be difficult due to jitter over increasingly large

distances, camera latency, and mis-registration issues. Kurz et al. (2007) created a reliable

method to track a laser pointer that started by creating a normal map of the environment. The

laser pointer could then be identified by comparing variance in the colour levels of ensuing

frames.

In a comparison of input devices for controlling a cursor on 2D surfaces, laser pointers were

found to perform worse than using a SmartBoard, a mouse, and relative mapping (Myers et

al. 2002). Relative mapping is a technique in which interactions on a larger surface are made

on a smaller device, such as a tablet, and are mapped to the larger surface in real time. This

allows a user to remain in place whilst interacting with a far larger display without having to

use distal pointing input. This makes relative mapping faster than interacting with the larger

18

surface directly in unimanual tasks, but not in bimanual ones (Forlines et al. 2007). Problems

with relative mapping are the slower deceleration experienced in virtual pointing, and a

reluctance of users to take advantage of virtually assisted targeting (Graham & MacKenzie

1996). Ninja Cursors (Kobayashi & Igarashi 2008) offers a partial solution to some of these

problems by placing several cursors on the environment which move in unison, decreasing

the distance a single cursor has to cover.

AR allows computer applications to more closely work with people to improve their work

processes. For example, the HandSCAPE (Lee et al. 2000) turns a measuring tape into an

input device to improve box packing into delivery vehicles. Users measure the three

dimensions of a box (length, width and height) and the system automatically stores these.

Once all boxes are measured, it arranges them in size and in order to be delivered in the most

optimal arrangement to fit in the vehicle. The ordering of the boxes is then projected to the

shipping workers.

Schwerdtfeger and Klinker (2008) compared several different methods of highlighting the

location of certain stored objects using a HMD to improve order picking tasks. The methods

examined were framing the target, drawing an arrow to the target, and creating a tunnel to the

target. These were demonstrated to improve the time taken to complete order picking tasks,

with framing and subtle tunnelling proving the most effective.

A Tangible User Interface (TUI) is one in which real-world objects are used to control a

variety of functions. Whilst the computer mouse is a common TUI device, TUIs can take

many forms. For example, simple blocks and cubes have been used as a TUI (FitzMaurice et

al. 1995). The IncreTable (Leitner et al. 2008) combines TUIs with projected images to create

a mixed reality game that operates in real time. The virtual objects are controlled with an

Anoto pen, and real objects are tracked with a camera. The system allows people to take a

virtual car, steer it up the incline of a real book, and perform a jump over the other side.

LEGO OASIS (Ziola et al. 2010) is a projector-camera system in which individual Lego

blocks are treated as projection surfaces mirrored by virtual objects. As Lego blocks are

combined, the nature and properties of the virtual block objects are also changed.

The Tango (Kry & Pai 2008) is a TUI with an accelerometer and 256 pressure sensors where

hand pressure is used as input for 3D interaction. A system using a stylus has been designed

to draw projected surgical notes on patients for use in surgery to avoid drawing on the

19

patient's skin in ink (Seo et al. 2007). Zaeh and Vogl (2006) developed an AR application

that allowed engineers to draw paths to steer robots on assembly lines in factories, instead of

having to type the paths into a computer.

Billiards cues and balls have been made SAR input devices to help teach beginners how to

take shots (Suganuma et al. 2008). The orientation and position of the cue is tracked from

above, and lines indicating shots that would sink the ball are projected onto the surface of the

table. The level of accuracy needed for tracking the cue proved a vast technical challenge.

Augmented foam sculpting is a SAR application in which cuts from a real piece of foam are

mirrored by cuts from a virtual model. Guidelines can be projected onto the surface of the

foam to assist in the cutting of foam to assist the user (Marner 2013). The system also

allowed for texturing the models at the same time. The benefit of this is a workable 3D model

is created identical to the foam sculpture being make, reducing the amount of work involved

in prototyping.

AR has also been used to prevent real-world collisions between actors. Oda & Feiner (2009)

created a multi-user hand-held AR game in which physical player interference caused by

players coming into physical contact with each other has been removed by transforming the

virtual locations of other players the physically closer they get, so that it led to a greater

distance to other players than other methods and reduced the game time as well.

20

2.5. Fitts' Law

With such a variety of possible input devices, a quantitative method of comparing their

performance is useful. Fitts' Law is a mathematical formula that can be used to compare the

difficulty in performing pointing tasks for a given input device, relative to the size of the

target and the variability of speed with which a user moves (Fitts 1954). Fitts derived the law

from the Weber Fraction and in a series of experiments was able to demonstrate its

applicability. The most commonly used form of Fitts' Law is given by:

(1)

where MT is the mean time to complete the task, a and b are regression constants such that a

is the start/stop time of the device and b is the inverse of the speed of the device, D is the

distance from the starting point to the centre of the target, and W is the width of the target.

The logarithmic term is called ID: the Index of Difficulty, representing how difficult it is to

complete the task.

There are several different variations on Fitts' Law. For example, Fitts' original function, in

the modern form in terms of MT, is given as:

(2)

where A is the amplitude of movement. The largest problem with this version is that it allows

for a negative, hence illogical, negative ID if the target was not in line with the direction of

movement. To overcome this issue, the Welford (3) and Shannon (4) formulations were

derived (MacKenzie & Buxton 1992).

(3)

(4)

The Shannon formulation was adopted as part of the ISO 9241 standard for measuring the

performance of non-keyboard input devices (MacKenzie et al. 2001). This was later

superseded by the modern form given in (1) where amplitude was replaced by distance.

MacKenzie and Buxton (1992) derived two more variations on Fitts' Law which in their

experiments improved the fit and reduced errors:

21

(5)

(6)

where H is the height of the movement, and W' is measured along the angle of approach.

All of the above formulae have been demonstrated to not account for the angle dependency in

pointing tasks. Kopper et al. (2010) suggest a new formula which takes the angle into

account:

(7)

.

where

.

and

such that DP is the distance perpendicular from the user to the surface, and k is a regression

constant. In their experiments they found that k fit best at 3.14.

Another formulation of the Steering Law containing a term for the angle of movement was

developed by Appert et al. (2008) and is given by:

(8)

where W is the width of the target and H is the height of the target. The inclusion of the

arbitrary constant term 0.6 in (8) increases the risk of overfitting the model for the data. Yang

and Xu (2010) developed the following formula which does not include such an arbitrary

term:

(9)

Whilst Yang and Xu's formula does not include arbitrary terms, it does assume a uniform

distribution of hits which other work has demonstrated to be inconsistent. Zhang et al. (2012)

developed the following formula which does not assume a uniform distribution whilst still

taking the movement angle into account:

22

(10)

such that 0 < ω < 1, and ω = c1 + c0 cos2θ, such that θ is the angle of movement of the

pointing task.

Fitts' Law is not without limitations. As movements exceed ≈40cm, Fitts' Law is more

sensitive to increases in the Index of Difficulty, and full limb and body movements are less

accurate for people to perform (Faconti & Massink 2007). The direction of movement also

has an effect on the difficulty of a pointing task, with vertical movements generally easier to

perform than horizontal ones and moving left-to-right providing a different result to right-to-

left. It has also been demonstrated to not account for the cognitive load of using a given

interface (Kabbash et al. 1994). Allowing for these constraints, Fitts' Law is a reliable and

well used method to compare non-keyboard input devices.

23

2.6. Steering Law

Differential calculus was used to extend Fitts' Law to two-dimensional steering tasks by

Accot & Zhai in 1997. Through user studies they conducted, the Accot-Zhai Steering Law as

it became known was confirmed empirically. They gave the Law as:

(11)

where C is the path, s is the abscissa along the path C, and W(s) is width the of the path at s,

such that the integral term represents the Index of Difficulty.

Accot and Zhai developed a Euclidean formulation of the Steering Law in 2003 (Grossman &

Balakrishnan 2005) for rectangular movement, given by:

(12)

where A is the amplitude of the movement, W is the width of the tunnel, H is the height of the

tunnel, and η is an empirically determined constant. The constant η is used to weight vertical

and horizontal movements differently following the observation that direction affects the

difficulty in completing the task.

With greater awareness of the effect that angle had on steering tasks, Grossman and

Balakrishnan (2005) proposed a probabilistic Steering Law capable of accounting for the

angle:

(13)

where R is a region defined by the target based on a spread of hits S, such that the universal

function F mapping the probability of hitting a target represents the Index of Difficulty.

To increase the generalisation of the Steering Law, Pastel (2006) investigated steering around

a 90º corner. The formula he derived is:

(14)

where c is a regression constant, IDS is the Index of Difficulty of the steering task on

approach to the corner, and IDF is the Index of Difficulty of a pointing task to the destination.

24

Pastel built on earlier reasoning by Ahlström that a steering task around a 90º corner is

comprised of a steering task to the corner, and a simpler task to the destination.

Zhou and Ren (2010) investigated the effect that bias towards speed or accuracy has on the

Index of Difficulty and Mean Time to Complete of steering tasks. They confirmed that the

faster a person attempts a steering task, the less correlation there exists between ID and MT.

They derived the following formula:

(15)

where A is the amplitude of movement and SD is the standard deviation of sampled points.

There are upper bound limits to the path width that can be modelled by the Steering Law

(Accot & Zhai 1997). Increasing the width of the tunnel too far in relation to its length breaks

the model's applicability in evaluating steering tasks. Scale also affects the steering tasks

being performed, with experiments by Accot and Zhai (2001) have shown that steering tasks

are optimal around A4 - A5 in size, consistent with a similar effect noticed in Fitts' Law that

as more arm movements are included the harder it is perform steering tasks accurately.

Other metrics exist that can be used to compare input methods quantitatively besides Fitts'

Law and the Steering Law. MacKenzie et al. (2001) list the following:

target re-entry

task axis crossing

movement direction change

orthogonal direction change

movement variability

movement error

movement offset

However not all of those are applicable to all input devices that can perform pointing and

steering tasks. Another metric they list is throughput, defined as:

(16 )

25

where We = 4.133SDx such that SDx is the standard deviation of selected coordinates

measured along the axis of approach to the target. The logarithmic term is IDe, the Index of

Difficulty of the steering task.

26

2.7. Mahl-Sticks

Figure 2.1: A mahl-stick in use. The left hand is holding the stick against wall, providing a prop that the

drawing hand can rest upon for added stability and to reduce arm fatigue.

A mahl-stick (also spelled maulstick, mahl-stick and mahl stick) is a traditional tool in

painting to support the brush hand. Mahl-sticks are typically around 1 metre in length,

cylindrical and padded on one end, although they are often made to any size and shape for the

personal preferences of the painter. When using a mahl-stick the painting surface is held

vertically, usually placed on an easel or fixed to the wall in some manner. The mahl-stick is

held by one hand on the far end; the painter can also hold a pot of paint or a palette in this

hand. The padded end is rested onto the easel, painting surface or wall - anywhere stable.

This end is sometimes wrapped in cloth or chamois to prevent or reduce damage to the

surface it is rested upon and to increase grip with the surface.

The brush hand rests on the top of the mahl-stick, which immediately provides the brush hand

more stability. Lines of any length are drawn by movements of both the brush hand and the

hand holding the mahl-stick. This motion effectively transfers control of the brush hand to

movements of the wrists and fingers, rather than movements of the shoulder, elbow, wrist and

fingers. As observed in analysis of pointing (Faconti & Massink 2007) and steering (Accot &

Zhai 2001) tasks, greater arm movements decrease the accuracy of the tasks performed.

Brush strokes are made downwards where possible. If left-handed, horizontal strokes are

usually made right to left, and if right-handed from left to right. Another important benefit

27

that a mahl-stick provides is that the fatigue of the brush-hand is considerably lessened.

Strokes are usually made at head-height. If possible the canvas is moved and rotated to ensure

that strokes are made in this manner.

My knowledge of mahl-sticks was obtained from my formal training in the Certificate II and

Certificate III of Signwriting at the Gilles Plains Institute of TAFE, South Australia. There is

scant information regarding the practice of using mahl-sticks in literature.

28

2.8. Summary

The existing literature on SAR applications reveals there is a wide range of potential future

applications for this technology. Artistic applications with a focus on drawing are certainly

some of them. The existing literature also reveals that simple tools and props are a preferred

method of interaction as they can take advantage of some of the innate benefits that SAR

technology offers whilst providing affordance to the functionality of tools already understood

by users. A mahl-stick is such a simple tool that has been in use for centuries. An objective,

qualitative method for comparing its effectiveness also exists in the Fitts' and Steering Laws.

The next chapter discusses the approach taken to compare the performance of pointing and

steering tasks with and without a mahl-stick.

29

3. RESEARCH METHOD

In this chapter the main experimental focus of this thesis is detailed: how the use of a mahl-

stick affects the performance of pointing and steering tasks in simple SAR drawing

applications. Two separate user studies were devised to test the use of mahl-sticks for these

tasks. The first study examines the effect of a mahl-stick in performing pointing tasks, and is

described in Section 3.1. The second study examines the effect of a mahl-stick in performing

steering tasks and is described in Section 3.2. Both studies were conducted in the same

environment using the same mahl-stick and stylus. This is all described in Section 3.3.

For practical reasons, both user studies were conducted simultaneously, and so the results and

their analysis is detailed in Chapter 4.

30

3.1. Pointing Task Study Methodology

Pointing is one of the simplest techniques available to interact with computer systems. In this

section the methodology used to evaluate the effect of a mahl-stick in performing pointing

tasks is described. This begins by stating the goal of the user study, followed by my

hypothesis of the results. Finally the design of the experiment is described in sub-section

3.1.3. As previously stated, the results and analysis from both user studies are detailed in

Chapter 4.

3.1.1. Goal

As discussed in Chapter 2, pointing tasks are a fundamental interaction technique in computer

systems and are a useful way to compare different interaction techniques. Examples of

pointing tasks in standard desktop computer usage include mouse clicking on desktop icons

and pressing keys on a keyboard. As a ubiquitous and common task well already understood

by computer users, designers of SAR systems may consider including pointing actions into

the systems they design.

Many extant SAR systems include pointing tasks (such as Dynamic Shader Lamps by

Bandyopadhyay et al. 2001, and the Build My Kitchen system by Marner 2013). Traditional

GUIs have been embedded into the real environment in a variety of ways (Feiner et al. 1993)

and many of the interactions within such systems is comprised of pointing tasks. SAR brings

the virtual world into the real-world through projection technology and through this offers a

wide variety of interaction methods.

A goal of this user study is to evaluate the effect that a mahl-stick has in facilitating the

performance of pointing tasks,. Changing the way in which users carry out tasks is one

method used in research to improve task performance. For example Jota et al. (2010)

managed to improve the accuracy of distal laser pointing simply by changing the way users

held a laser pointer. In the same vein this study will look at the effect on pointing tasks by

simply comparing the difference between the tasks performed with a mahl-stick and the same

number of tasks performed without.

31

3.1.2. Hypothesis

My hypothesis for this user study was that the mahl-stick could offer no improvement to the

user in completing simple pointing tasks. When painting and signwriting the mahl-stick

offers its advantages in supporting the painter to make precise and even brush strokes,

however such activities are more comparable to steering tasks. A pointing task does not

require such accuracy over the length of travel and I hypothesised that such tasks would not

leverage the stability a mahl-stick provides and that the stick itself could restrict the field of

vision of the user.

3.1.3. Pointing Task Study Design

In keeping the practice of using a mahl-stick, the experiment was conducted on a vertical

wall. The design of this user study was based on the large-scale pointing study on a

whiteboard performed by Faconti and Massink in 2007. To keep within the space constraints

in using a mahl-stick and to reduce body movement, the scale of this experiment was reduced

to be within a 300 × 300mm area in which the pointing targets were placed.

Ultimately it was decided that users would perform four blocks of pointing tasks: two with a

mahl-stick and two without, interleaved. Whether the user started with or without the mahl-

stick was randomly determined. By interleaving the blocks it would be possible to compare

the improvement both with and without the mahl-stick, and randomising whether to start with

or without the stick would reduce the bias of the learning effect in the final results.

In Faconti & Massink's experiments, participants were asked to travel from one of 5 targets to

another randomly determined target. Four of the targets were placed in each of the four

corners of a rectangular whiteboard, and the fifth was placed directly in the centre of board.

This user study adapted this setup to instead be a 3 × 3 array of nine circles arranged linearly

and equidistantly, as in Figure 3.1 below. Each circle was 12mm in radius and spaced 135mm

apart. The addition of 4 circles along the edges served to help keep the lengths of the travel

between targets to between those for when I was first instructed in using a mahl-stick whilst

still allowing longer distances. The dimensions were chosen as they would keep the area to

within the 300 × 300mm size constraint. These targets were projected onto a wall using a

virtual rear projection technique, as described in greater detail in Section 3.3 below.

32

Figure 3.1: Arrangement and size of the projected targets in the Pointing Task study

The height and placement of the task area was made adjustable; it could be raised or lowered

to accommodate the height of each participant with the goal to place the central circle near

eye-level when standing. By default this height was approximately 1600mm. This was to

match with the typical use case of a mahl-stick where the canvas is raised or lowered where

possible so that the painter is working at eye-level.

Figure 3.2: The preferred height of the task circles in relation to a participants' height.

The experiment made use of a stylus which used an infrared optical tracking technique. The

design and implementation of the stylus is detailed in Section 3.3 below. The system tracked

the movement of the stylus through real-world space.

33

Participants were given a brief description of mahl-sticks, shown how to use them, and were

given the opportunity to practice using the stick and the stylus before the experiment began.

Only once the participant indicated they were ready did the user study begin.

a)

b).

Figure 3.3: The participant performing the task without the stick (a) and with the stick (b)

All of the 9 points were coloured in light blue (RGB: 127, 127, 255) by default. The

designated target was coloured yellow (RGB: 255, 255, 0). This provided a stark visual

contrast for the target and was clear in the environment. All 9 targets were projected at all

times. After completing each block the circles would turn grey for 5 seconds (RGB: 127, 127,

127), indicating to the participant that the stage had been completed.

Each block was comprised of 35 tasks; each task being defined by an origin point and a

destination point. The path for each task was restricted to only vertical, horizontal, and 45º

angled paths so as to remain consistent with the Steering Task Study, as described in the next

section. There were a total of 56 valid paths; the paths were randomly chosen for each block

for each participant and no path would repeat in one block. This would provide a wide a

range of paths with which to compare across all results. Originally it was planned for

participants to complete all 56 paths each block. However initial testing revealed that it was

too fatiguing on the participants arms when performed in succession with the Steering Task

study, and so the number was reduced to 35 which still resulted in a large number of paths to

compare blocks against.

For each task one circle would be highlighted at random (Figure 3.4, a). When the user

positioned the tip of the stylus over that target, another circle along a valid non-repeating path

would be highlighted at random (Figure 3.4, b). These paths were pre-programmed into an

34

array and selected at random, to ensure that no origin point could be selected that had already

exhausted all possible paths.

a)

b).

Figure 3.4: a) An origin point chosen at random. b) Once the participant has placed the tip of the stylus over the

origin point, a destination point is displayed.

After each task is completed another origin point was chosen at random; it may or may not

have been the same as the destination point. To prevent any possible confusion to the

participants, after completing a task the next origin point was not highlighted until the user

removed the tip of the stylus from the wall.

For each task in each block the system recorded several items of data: the participant's

number; whether or not they used a mahl-stick; the block number (starting at 0); the origin

point for the task; destination point; the time it took to complete the task; and, how far they

were from the centre of the destination point when they completed the task recorded in

millimetres. This distance can be used to rate the average accuracy in completing the pointing

tasks as the closer they were to the absolute centre of the target the more accurate they were.

This measurement could not be used to rate the absolute accuracy of each individual task as

the stylus' position jittered in the system to within approximately 2mm from its actual

position. This jitter was accounted for in the design (see Section 3.3 below) of the

experiment, and would average out in a large enough population. This information is enough

to evaluate the mahl-stick in terms of Fitts' Law as well as obtain data on a learning effect.

After completing all four blocks, the user was asked to fill in a short survey asking their age,

sex, and past experience with a mahl-stick. They were also asked to rate from 1 to 5 how

easy, accurate, and fast they felt they were in completing the tasks with and without a mahl-

35

stick. Finally they were asked to rate their preference in using a mahl-stick in performing

pointing tasks.

Also during the task participants were assessed on which technique they used to hold the

stick and how they were coping with arm fatigue.

36

3.2. Steering Task Study Methodology

Steering is another simple technique available to interact with computer systems. In this

section the methodology used to evaluate the effect of a mahl-stick in performing steering

tasks is described. This is begun by stating the goal of the user study followed by my

hypothesis of the results. Finally the design of the experiment is described in sub-section

3.2.3. As previously stated the results and analysis are detailed in Chapter 4.

3.2.1. Goal

As discussed in Chapter 2, steering tasks are another fundamental interaction technique in

computer systems. A common example in desktop computing of a steering task is navigating

a menu in a program. Another example of a steering task is to draw a line through a specific

tunnel.

Drawing lines is used extensively in creative and design processes and SAR has been

demonstrated capable of benefiting existing workflows. For example, the Digital Airbrushing

and Augmented Foam Sculpting systems contain drawing or steering tasks (Marner 2013).

Drawing tasks are also used in entertainment; for example the IncreTable (Leitner et al. 2008)

is a mixed-reality game that demonstrates SAR applications in this field.

A goal of this user study is to evaluate the effect that a mahl-stick has in facilitating the

performance of steering tasks to both draw lines and navigate systems. As far as I can

determine this is the first research of this type on mahl-sticks.

3.2.2. Hypothesis

My hypothesis for this user study was that the mahl-stick will offer an improvement to the

accuracy of simple drawing tasks at the expense of speed. In painting and signwriting the

mahl-stick offers advantages by supporting the painter's brush hand to make precise and even

strokes. However doing so is noticeably slower than drawing freehand. These benefits should

be able to translate to a mixed reality drawing context.

37

3.2.3. Steering Task Study Design

This user study was designed to be as similar to the Pointing task user study as possible. The

same arrangement, size and spacing of targets was retained as in Figure 3.5 below. Users

were asked to perform four blocks of steering tasks: two with a mahl-stick and two without,

and like in the Pointing task study they were interleaved. It was randomly determined

whether to start with or without the mahl-stick.

Figure 3.5: Arrangement and size of the projected targets in the Pointing Task study

This study had the same setup and environmental design as the pointing task user study: two

projectors overlapping their projections to create a virtual rear projected display. The infrared

optically-tracked stylus and mahl-stick were retained and the tracking system was no

different. The height of the circles was also adjustable to suit the height of the participant.

Each block was comprised of 35 tasks, comprised of the 56 randomly selected path

orientations as in the pointing study: constricted to those routes that are either vertical,

horizontal, or at a 45º angle. Angle has been shown to have an important effect in completing

steering tasks (Grossman and Balakrishnan 2005). Direction is important when using a mahl-

stick: downward strokes are generally considered the easiest to make, and horizontally it is

easier to make strokes in the same direction as you are handed. By limiting the paths to those

angles it would be easier to analyse path directions if something were significant was found.

Restricting to those angles also allows potential comparisons to be made regarding the effect

length played in performing those tasks.

38

For each task one circle would be highlighted at random (Figure 3.6, a). When the user

positioned the tip of the stylus over that target, another circle along a valid non-repeating path

would be highlighted at random as well as a path as wide as the circle towards it (24mm)

(Figure 3.6, b). These paths were pre-programmed into an array and selected at random, to

ensure that no origin point could be selected that had already exhausted all possible paths.

a)

b).

Figure 3.6: a) An origin point chosen at random. b) Once the participant has placed the tip of the stylus over the

origin point, a destination point is displayed.

The goal of each task is to draw a line across the wall from the origin target to the destination

target without leaving the path. If the stylus left the path it would be recorded as an error, and

all the circles would flash red for half a second before starting the next task. After completing

a task the next origin point was not highlighted until the user removed the tip of the stylus

from the wall.

For each task in each block the system recorded several items of data: the participant's

number; whether or not they used the mahl-stick; the block number (starting at 0); the origin

point for the task; destination point; the time it took to complete the task, and; the distance

they travelled from the origin to the edge of the destination, measured as the sum of straight

line segments every 20 microseconds. This distance can be used to rate the average accuracy

in completing the pointing tasks as the lower it is as the straighter the path was, the more

accurate they were. Because of jitter with the tracking system, a tolerance buffer was placed

around the path to ensure that users were not marked as erring even though they were still

within the tunnel.

39

a)

b).

Figure 3.7: A participant in the steering task study, with a mahl-stick (a) and without (b).

After completing all four blocks, the user was asked to fill in a short survey asking their age,

sex, and past experience with a mahl-stick. They were also asked to rate from 1 to 5 how

easy, accurate, and fast they felt they were in completing the tasks with and without a mahl-

stick. Finally they were asked to rate their preference in using a mahl-stick in performing

steering tasks.

During the task participants were assessed on which technique they used to hold the stick and

how they were coping with arm fatigue.

Originally the paths for the task were much smaller: the radius of the circles was 7.5mm and

the width of the tunnels was 15mm. After running an initial participant it became clear that

this was too difficult for members of the general public to perform. For the need to obtain

actual data which was not almost entirely a sequence of errors, the size of the circles and the

width of the tunnels was increased to 24mm.

40

3.3. User Study Environment

The same setup was used for both the Pointing and Steering task user studies, in the same

location. The same mahl-stick and stylus were also used. This section describes the theory

and processes used to calibrate the system to run the user studies, and the details describing

the construction of the stylus and mahl-stick. The system used the SAR software modules of

the Wearable Computer Lab at the University of South Australia, with several modifications

made to adapt to the system to what was required. The first section describes the projection

system used and the second section describes the tracking system. The stylus and mahl-stick

are then described in detail. The module created to combine these elements and manage the

user study is then described.

3.3.1. Projection System

SAR uses projectors to alter the appearance of objects in real-time. One of the more

immediate problems when using projectors is the risk of shadow occlusion. If one object is

between the projector and the surface it is projecting onto, a shadow is cast and that part of

the image is lost. This was an issue in the pointing and steering task studies: as the user

would be performing tasks on a projected wall, no matter where a projector is positioned

behind the participant there will be always be shadow occlusion. This is demonstrated in

Figure 3.8 below. If it is placed at an angle oblique enough so that the participant's body did

not cause occlusion, the image would be so distorted that accuracy and image quality would

be sacrificed. When the stylus came in contact with the surface, the image would be occluded

anyway.

Figure 3.8: Use of a single projector would have left a large occluded area where the circles would not be

visible without the participant moving out the way, greatly hindering performance of pointing and steering tasks

41

To overcome this issue a virtual rear projection technique was used, as described by Summet

et al. (2005). This uses several projectors placed at different oblique angles away from the

surface of a wall, which are calibrated to project the same image onto the same section of

wall. This is done so that if a person occludes the projection from one projector, the image

can still be seen as the other projections are not being occluded.

This is demonstrated in Figure 3.9: (a) shows the occlusion from a projector to the left, and

(b) the occlusion from a projector to the right. As described above, if only one of projector is

used, part of the image is occluded. However when two projectors are projecting

simultaneously from different locations (c) the resulting image is still clear as if the user is

occluding one projector, the image from another projector is very likely still clear. This made

it unlikely that any one target could be fully occluded.

a)

b).

c)

Figure 3.9: The images from left (a) and right (b) projectors individually showing the occlusion (the greyed out

area), and the combined image from both projectors (c).

There are several reasons why this projection system was chosen over other methods. As

stated, if only a single projector was used shadow occlusion would have been a significant

issue. No matter where it was placed the participant would occlude it at some point when

performing the user study which would have introduced errors into the results: users would

have been forced to either guess where the targets were or move around considerably

affecting the resulting time to complete each task.

42

The use of a Smart Board was also inappropriate this purpose. Whilst this would have solved

all issues with shadow occlusion, a mahl-stick is used to support the drawing hand.

Depending on the force the participant placed on the stick, damage to the board surface could

have occurred. The goal of this study was to examine mahl-sticks in a SAR context: a rear

projection technique such as a Smart Board was not a SAR-based solution to the occlusion

issue.

Two projectors were deemed sufficient to project onto the wall. These were positioned 3

metres above floor level, 2.5 metres away from the wall. Both projectors were placed off-

from-centre to the targeted projection area to help reduce shadow occlusion. A colour-

banding technique was used to calibrate both projectors automatically. A digital camera was

set back from the wall positioned so that the full images from both projectors were visible. A

sequence of black and white bands were projected through each projector. The camera

captured these bands and from the resulting data the intrinsic and extrinsic properties of each

projector were calculated.

Figure 3.10: Even though his hand occludes the light from one projector, the path can still be seen as the other

projector is not occluded.

An 1135 × 976mm rectangular level area was marked out onto the wall. A crosshair was

moved through the projected image to record the corners of this area, relative to the

projections. Doing this allowed the projections to be cropped and keystoned to align to this

area, scaled to the default OpenGL floating point scale of 0.0 to 1.0 in both the x and y axes.

It was trivially simple to change this scale to 0 - 1135 in the x-axis and 0 - 976 in the y-axis to

match the real-world scale, so that the size of image elements could be set to millimetres and

that the units could also match the tracking system, described in the next section.

43

3.3.2. Tracking System

An infrared optical tracking system was chosen to track the stylus. Infrared tracking was

readily available in the Wearable Computer Lab and supported in the pre-existing software

modules. From an implementation standpoint a robust optical tracking solution already had

much of the groundwork for it laid and it would have been less effort to get it to work in the

modules. As described in Chapter 2, Section 3, optical tracking is fast, has high update rates

and is unaffected by surrounding electromagnetic interference. Optical tracking is also widely

used in SAR already. However it suffers from line-of-sight issues as the optical markers need

to be visible from the tracking cameras.

Other tracking solutions were dismissed. The laboratory environment the user study was run

in has many electronic devices situated through it and man cables running through the floor,

ceiling and walls. Magnetic trackers could not be used in such an environment as the

potential for electromagnetic interference would have been too great, and obtaining another

environment to run such a study would have been too difficult. Acoustic trackers are slower

than optical trackers but still suffer from line-of-sight issues, making them an inferior option.

Devising a mechanical tracking solution could potentially have been expensive and a difficult

engineering problem: devising a hand-held stylus that could be moved easily, be held by

people comfortably as well as used in conjunction with a mahl-stick without impeding the

participant's sight.

Combining optical tracking with inertial tracking was considered, as inertial trackers have

high update rates and are unaffected by the line-of-sight issues that impede optical tracking.

By embedding a 6 degree-of-freedom accelerometer and gyroscope (sometimes called an

Inertial Measurement Unit), which tracks movement along 3 axes and the rotation forces

around each, inside the stylus it would have been possible to keep track of the stylus in

situations where line-of-sight was broken. But this leads to an inevitable problem - when

optical and inertial trackers give conflicting tracking information how does one know which

one is reporting correct information, if at all? Also out of concerns for the physical

construction of the stylus (described in the next sub-section 3.3.3) the idea of combining

optical and inertial tracking was abandoned.

The OptiTrack motion capture software created by NaturalPoint was used to register the

position and angle of the stylus. Infrared retro-reflective tape was wrapped around 4 marker

44

spheres attached to the stylus. Five Flex 3 motion capture cameras were positioned around

the projection area, as shown in Figure 3.11 below: one placed to the left looking between the

participant and the wall, and four suspended from the ceiling look down towards the user.

Figure 3.11: The arrangement of Flex 3 cameras positioned around the projection area.

VRPN, a device-independent networking protocol to track peripherals in VR applications,

was used to communicate the tracked data from the OptiTrack system into the Wearable

Computer Lab's SAR modules. The points from OptiTrack were originally based around a

coordinate system set by the positions of the cameras. These camera space coordinates were

transformed to match the projector's coordinate system: the x-axis was transformed along the

length of the projection, the y-axis was transformed along the height of the projection, and the

z-axis was transformed so that it emerged outwards from the wall perpendicular towards the

participant. The scale of the coordinates changed to match the size of the projection plane,

with the origin (0, 0, 0) set to be the bottom left-hand corner of the wall.

Figure 3.12: The final coordinate system of both projectors and the camera space

45

3.3.3. Stylus

It was determined that a stylus was the optimal way to interact with the system in the pointing

and steering task studies. A stylus is already used as an input device in SAR systems. It is a

useful replacement for a brush in drawing tasks: a stylus is in many ways a simplification of a

brush. A stylus was created for the pointing and steering task studies.

As stated before in the 3.3.2. Tracking section, it was originally intended that the stylus

would make use of both optical tracking through infrared retro-reflective markers mounted

onto it, and inertial tracking through an embedded electronic 6 degree-of-freedom gyroscope

- accelerometer. Whilst being able to correctly determine the position of the stylus when the

inertial and optical tracking were providing conflicting accounts was one concern that

ultimately led to the abandonment of this idea, the resulting size that such a stylus would

have was another serious concern. The stylus had to be small enough to be held in any

participant's hands, whilst at the same time large enough to carry a microcontroller and

accelerometer. This would have resulted in a narrow stylus with a large and overweighted

protrusion to housing electronics at the rear. The 3D printer which would ultimately create

the stylus can only print in a volume 250 × 250 × 250mm. Initial trials of combining two

pieces of 3D printed objects together resulted in breaks when held roughly. It could not be

ensured that trial participants would be gentle with the stylus. With all these concerns the

stylus was simplified to only use optical tracking and was designed to be less than 250mm

long.

Figure 3.13: The final stylus used in the trial

46

The main part of the stylus was created 15mm in diameter and 200mm in length. The tip

extended 18mm and ended in a 3mm rounded cap. Four 14mm spheres were attached at

length from the body of the stylus: one directly from the rear, two at different lengths at 90º

from each other near the back, and one a short distance from the centre. The four spheres

were covered in infrared retro-reflective tape to work with the OptiTrack system. The four

spheres were printed as part of the stylus, and not attached afterwards, to provide additional

rigidity.

3.3.4. Mahl-Stick

There is not set form, size or length a mahl-stick can take; some mahl-sticks can range from

the short to over 4 metres in length. They are usually made of wood though aluminium or any

other rigid and relatively light material can be used. They are often rounded in cross-section

so as to not cause discomfort as the user rests their wrist on a hard edge, though rectangular

and square mahl-sticks exist. Sometimes one end is capped in a protrusion or stopper of some

form, and this end can be wrapped in a cloth or chamois to protect the surface its resting on or

decrease the chance of slippage.

Without a standard form or size, for the purposes of this study a round 20mm diameter pine

mahl-stick 1200mm long was created. This was chosen as it is a length that would suit most

anybody drawing within a 300 × 300mm area, is lightweight and was affordable. One end

was capped in a 50mm diameter sphere, which was 3D printed, so as to not damage the wall

with a sharp edge of the wood. In turn this was wrapped in a piece of cloth to increase grip

with the wall.

Figure 3.14: The mahl-stick used in the user studies

47

4. ANALYSIS

19 participants took part in the combined pointing task and steering task user study, with

participants drawn from the Mawson Lakes campus of the University of South Australia and

the general public. The combined study took approximately 30 minutes to complete. After

completing both the pointing and steering task studies, participants were asked to answer a

questionnaire regarding their opinions of the results.

The pointing task and steering task studies were ran one after another. It was randomly

determined which study the participants started with. It was also randomly determined

whether they would start with using the mahl-stick or not. After completing each block of 5

tasks, a 5 second interval would occur where the stick was swapped in or out. After the first

four blocks were complete, users were given some time to rest if they needed before starting

the other study.

After the first participant was ran, changes were made to both the pointing and steering task

study. The radius of the circles was increased from 7.5mm to 12mm and the width of the

tunnels were made as wide as the new circles. This was done because the initial tunnel width,

15mm, was too narrow for members of the general public to complete. Moving the stylus

outside of the tunnel resulted in an error, and the error rate in the first steering study ran

around 90%. After the size of the elements was increased, no further alterations were made to

the study. Only the results from the remaining 18 participants were analysed.

The research question was to:

...evaluate the effect that a mahl-stick has when performing simple pointing and

steering tasks in Spatial Augmented Reality applications when using a stylus.

As part of this two sub-questions were asked:

To compare how novice and experienced users compare with the use of a mahl-stick

in pointing and steering tasks, and;

what learning effect is observable in using a mahl-stick.

48

Participants were between the ages of 24 and 48. 16 were male and 3 were female. All

participants were right-handed. Only one participant (19) had used a mahl-stick prior to the

commencement of the study. As I was unable to get more participants that had used a mahl-

stick prior to the study, I was unable to obtain enough data to answer the first sub-question.

Over the course of the user study, from observation of the participants and conversation with

them, it became clear that fatigue had a large effect on the performance of pointing and

steering tasks. However as fatigue was not asked about in the survey, the participants

themselves did not write down the level of fatigue they felt they experienced.

The analysis of results is broken into 3 parts: the analysis of the pointing task study results in

section 4.1, the analysis of the steering task study in section 4.2, and an analysis of the

surveys in section 4.3.

49

4.1. Pointing Task User Study

I hypothesised that a mahl-stick could offer no real improvement to accuracy in performing

pointing tasks, and that it would reduce arm fatigue at the cost of making the pointing tasks

slower to perform.

The data recorded for the pointing task study was:

the participant number,

the block number,

whether the mahl-stick was used or not,

the origin point for each task,

the destination point for each task,

the time taken to complete each task, and

the distance from the centre of the destination point.

The origin and destination points allows one to determine the length of each path from origin

to destination, and the direction in which the travel occurred. The distance from the centre of

the destination target provides an indication of the accuracy of the pointing tasks; the greater

it is the less precise they were in selecting the target.

The data for the results is tabled in terms of time in milliseconds, distance from the centre of

the destination point in millimetres, and the total speed for which the task was completed

(calculated from the length of the path for each task divided by the time taken to complete)

measured in millimetres per millisecond, or metres per second. All results were trimmed to be

within 3 standard deviations of the mean, as per the standard practice in Computer Science.

50

Time to Complete Tasks (ms)

1st Block 2nd Block

With Mahl-stick Without Mahl-stick With Mahl-stick Without Mahl-stick

Mean ( ) SD (σ) Mean ( ) SD (σ) Mean ( ) SD (σ) Mean ( ) SD (σ)

2 1428.286 331.598 1084.324 309.115 1292.088 281.678 944.441 222.654 3 844.382 261.594 776.735 212.854 764.941 151.030 711.914 82.224 4 1443.543 499.571 1161.706 486.100 1182.971 243.467 984.352 327.205

5 889.829 172.797 793.600 103.469 802.257 103.268 716.486 65.7407 6 928.303 1413.740 816.576 235.357 859.177 245.852 779.212 147.100

7 1130.629 232.734 1037.588 584.103 1067.500 334.131 1076.714 149.868 8 1068.714 242.178 884.152 281.442 913.485 548.814 790.265 176.229

9 713.265 215.157 696.515 246.127 654.500 197.592 646.353 102.163 10 824.559 170.685 886.235 568.080 909.618 402.890 771.177 182.635

11 780.971 200.275 861.114 260.098 863.735 277.758 826.823 329.517 12 896.257 227.646 916.735 565.147 968.571 397.417 686.765 196.434 13 854.618 211.826 783.794 189.721 728.486 444.450 919.879 105.824

14 1025.471 248.511 952.235 404.607 1003.118 240.338 846.294 153.696 15 1349.559 596.214 994.382 363.534 1166.441 340.281 765.559 192.237

16 1049.971 349.480 817.029 102.768 937.697 159.293 823.229 118.821 17 1531.618 941.112 892.471 182.214 1341.086 376.428 1007.000 208.529

18 737.371 166.772 683.677 161.239 648.412 112.07 707.114 606.433 19 821.441 152.457 970.857 120.255 759.457 121.189 760.257 142.588 ALL 1066.551 581.164 916.840 362.186 976.975 354.604 826.130 256.806

Distance from Centre of Pointing Target (mm)

1st Block 2nd Block

With Mahl-stick Without Mahl-stick With Mahl-stick Without Mahl-stick

Mean ( ) SD (σ) Mean ( ) SD (σ) Mean ( ) SD (σ) Mean ( ) SD (σ)

2 4.457 3.475 3.588 2.289 3.882 2.904 4.324 2.529 3 8.206 4.482 7.353 4.132 10.412 4.736 8.343 4.665

4 8.486 4.736 8.824 5.035 9.171 5.079 10.265 4.937 5 6.286 4.322 5.800 2.743 5.286 3.553 7.686 3.279

6 7.818 4.426 7.424 4.341 9.088 4.943 8.485 3.909 7 6.857 4.074 5.118 3.311 8.824 4.965 5.143 3.549

8 6.343 3.531 6.848 3.733 8.212 4.378 5.265 2.944 9 10.382 4.467 7.121 4.759 10.029 4.649 7.118 4.098

10 8.647 4.125 7.559 3.615 7.294 4.120 5.618 3.369 11 9.382 4.979 7.457 3.921 7.324 3.956 6.765 3.483 12 9.343 4.556 11.471 3.936 10.543 4.931 8.529 4.619

13 8.618 4.996 7.294 4.253 7.200 4.549 8.061 3.917 14 8.441 5.121 7.176 4.530 7.853 4.740 9.029 5.042

15 9.676 4.371 7.676 3.936 8.176 4.232 9.794 4.821 16 8.059 5.396 8.206 4.444 8.152 4.684 6.486 3.899

17 8.029 5.621 7.618 4.649 9.514 5.299 6.529 4.473 18 9.571 4.698 8.147 4.178 8.853 5.057 10.114 4.741 19 8.382 4.230 6.743 3.791 6.543 3.868 8.143 4.532

ALL 8.119 4.720 7.314 4.320 8.243 4.776 7.414 4.367

51

Overall Speed to Complete Tasks (ms-1)

1st Block 2nd Block

With Mahl-stick Without Mahl-stick With Mahl-stick Without Mahl-stick

Mean ( ) SD (σ) Mean ( ) SD (σ) Mean ( ) SD (σ) Mean ( ) SD (σ)

2 0.139 0.043 0.181 0.063 0.143 0.040 0.216 0.069 3 0.242 0.081 0.260 0.083 0.246 0.071 0.276 0.086 4 0.148 0.048 0.160 0.047 0.168 0.054 0.204 0.058

5 0.229 0.066 0.251 0.069 0.224 0.057 0.257 0.089 6 0.212 0.082 0.240 0.077 0.236 0.072 0.265 0.081

7 0.177 0.050 0.192 0.063 0.186 0.059 0.193 0.068 8 0.183 0.059 0.223 0.065 0.211 0.067 0.250 0.086

9 0.271 0.086 0.289 0.102 0.295 0.090 0.305 0.097 10 0.218 0.061 0.256 0.105 0.224 0.078 0.266 0.080

11 0.243 0.080 0.244 0.075 0.241 0.077 0.254 0.096 12 0.219 0.061 0.245 0.104 0.207 0.085 0.295 0.088 13 0.240 0.065 0.247 0.069 0.260 0.072 0.217 0.079

14 0.207 0.052 0.219 0.077 0.207 0.064 0.223 0.074 15 0.153 0.052 0.197 0.064 0.159 0.036 0.257 0.088

16 0.184 0.066 0.254 0.084 0.215 0.064 0.244 0.080 17 0.137 0.050 0.227 0.079 0.152 0.047 0.185 0.057

18 0.266 0.082 0.273 0.089 0.291 0.087 0.286 0.104 19 0.250 0.076 0.253 0.087 0.248 0.075 0.265 0.084 ALL 0.204 0.077 0.231 0.085 0.212 0.078 0.248 0.087

Overall participants tended to become faster across the two blocks, both with and without the

stick. Performance was greater without the use of the mahl-stick than with it. There was not a

significant difference in the accuracy of pointing tasks. A repeated measures ANOVA was

performed on the Time to Complete Tasks, the Distance from the Centre of Pointing Target,

and the Overall Speed to Complete Tasks. No results of significance were detected (p > 0.05),

and no learning effect was observed in using the mahl-stick.

Below the Index of Difficulty for the four possible line lengths are calculated, using the

formulae of the Shannon IDS, Welford IDW, and Yang & Xu's IDYX formulations. The size of

the targets did not change from 24mm, only the distance between targets varied. HV1 and

HV2 are paths that existed either vertically or horizontally, one or two targets away

respectively. D1 and D2 are diagonal paths at 45º, one or targets away respectively as well.

The length L of the lines are given in millimetres.

L IDS IDW IDYX

HV1 135 2.728 2.615 1.392 HV2 270 3.615 3.555 2.087

D1 190.918 3.163 3.080 1.721 D2 381.838 4.080 4.036 2.484

52

4.2. Steering Task User Study

My hypothesis for this user study was that the mahl-stick will offer an improvement to the

accuracy of simple drawing tasks at the expense of speed. In painting and signwriting the

mahl-stick offers advantages by supporting the painter's brush hand to make precise and even

strokes, though it is noticeably slower than drawing freehand. These benefits should translate

to a mixed reality drawing context substituting a stylus for the brush.

The data recorded for the pointing task study was:

the participant number,

the block number,

whether the mahl-stick was used or not,

the origin point for each task,

the destination point for each task,

the time taken to complete each task,

the length of the line drawn, measured in line segments every 20ms, and

whether the task ended in error or not.

As with the pointing task study, the origin and destination points allows the length of the

ideal path and the direction to be calculated. The length of the line drawn was useful for

calculating the accuracy of the line drawn. The shorter the line drawn the more accurate it

had to be, and likewise the longer the line the less accurate.

The data for the results is tabled in terms of time in milliseconds, distance from the centre of

the destination point in millimetres, and the total speed for which the task was completed

(calculated from the length of the path for each task divided by the time taken to complete)

measured in millimetres per millisecond, or metres per second.

All results were trimmed to be within 3 standard deviations of the mean, as per the standard

practice in Computer Science. Due to occlusions in the tracking system, not all errors were

caused by users leaving the bounds of the tunnel. If the user held or moved the stylus in such

a way as to occlude one of the markers, the OptiTrack software lost track of the real-world

position of the stylus and assumed it was elsewhere. If this occurred whilst drawing a line, the

user study module would obtain the data and assume the stylus had left the bounds of the

53

tunnel, recording an error. Because it was impossible to state with certainty which errors were

user-caused and which weren't, all errors were removed from the analysis of results.

Time to Complete Tasks (ms)

1st Block 2nd Block

With Mahl-stick Without Mahl-stick With Mahl-stick Without Mahl-stick

Mean ( ) SD (σ) Mean ( ) SD (σ) Mean ( ) SD (σ) Mean ( ) SD (σ)

2 1302.739 275.838 1150.371 317.176 1770.364 613.084 1221.030 382.410 3 901.344 156.170 758.645 120.059 830.667 133.231 913.394 112.230 4 1581.833 353.802 1448.615 335.487 1478.400 341.619 1215.231 308.111

5 1496.576 436.753 1221.091 377.507 1378.571 322.489 1074.559 298.623 6 1320.645 319.513 800.879 143.357 1057.280 325.454 879.231 189.237

7 1713.259 564.150 1423.452 356.833 1284.030 328.239 1461.250 239.264 8 1528.435 548.967 1072.000 403.191 1571.583 420.711 1161.742 237.122

9 883.276 176.258 724.097 109.245 773.267 151.285 794.065 136.249 10 1187.040 208.816 1177.412 260.919 1332.368 327.546 1143.897 338.943 11 753.960 164.355 754.241 89.724 825.607 306.884 824.963 102.471

12 1665.929 532.312 844.136 136.248 1207.935 409.317 791.419 168.784 13 833.424 101.723 786.786 99.886 871.800 126.742 764.533 101.025

14 1565.567 344.554 1317.629 271.847 1022.656 307.092 1461.240 129.986 15 1333.462 270.804 1102.885 234.449 1465.281 256.199 1055.000 262.240

16 2904.045 765.364 1165.857 339.598 1336.273 939.710 916.333 122.821 17 3430.294 1673.309 1560.759 380.863 3547.136 1153.255 1441.517 610.513

18 868.903 139.350 774.382 100.224 749.417 154.898 886.433 84.411 19 2524.656 729.694 1143.697 433.485 1219.824 211.054 956.788 237.469 ALL 1497.259 859.701 1073.397 381.451 1338.072 707.945 1026.209 334.117

Length of Line Drawn (mm)

1st Block 2nd Block

With Mahl-stick Without Mahl-stick With Mahl-stick Without Mahl-stick

Mean ( ) SD (σ) Mean ( ) SD (σ) Mean ( ) SD (σ) Mean ( ) SD (σ)

2 1.232 0.150 1.202 0.078 1.286 0.203 1.233 0.125 3 1.226 0.099 1.239 0.111 1.199 0.141 1.221 0.111 4 1.288 0.192 1.241 0.135 1.445 0.224 1.338 0.353

5 1.246 0.109 1.374 0.185 1.353 0.150 1.249 0.121 6 1.368 0.284 1.242 0.092 1.307 0.197 1.281 0.146

7 1.243 0.140 1.311 0.188 1.189 0.069 1.226 0.081 8 1.452 0.293 1.343 0.204 1.231 0.129 1.245 0.104

9 1.250 0.116 1.226 0.097 1.267 0.169 1.179 0.092 10 1.265 0.112 1.208 0.131 1.259 0.154 1.206 0.098 11 1.211 0.097 1.217 0.114 1.275 0.160 1.216 0.111

12 1.350 0.202 1.263 0.100 1.232 0.118 1.285 0.166 13 1.238 0.147 1.314 0.185 1.246 0.186 1.213 0.097

14 1.254 0.179 1.278 0.141 1.238 0.162 1.288 0.114 15 1.336 0.155 1.345 0.180 1.379 0.228 1.299 0.204

16 1.938 0.714 1.220 0.124 1.273 0.527 1.260 0.110 17 1.772 0.508 1.328 0.249 1.844 0.397 1.361 0.231

18 1.284 0.238 1.261 0.175 1.302 0.134 1.280 0.230 19 1.587 0.427 1.226 0.146 1.253 0.169 1.219 0.132 ALL 1.348 0.319 1.268 0.158 1.310 0.253 1.248 0.161

54

Speed to Complete Tasks (ms-1)

1st Block 2nd Block

With Mahl-stick Without Mahl-stick With Mahl-stick Without Mahl-stick

Mean ( ) SD (σ) Mean ( ) SD (σ) Mean ( ) SD (σ) Mean ( ) SD (σ)

2 0.146 0.029 0.174 0.036 0.124 0.028 0.168 0.030 3 0.232 0.064 0.242 0.061 0.259 0.049 0.221 0.065 4 0.116 0.021 0.147 0.036 0.153 0.042 0.165 0.026

5 0.137 0.031 0.186 0.034 0.161 0.032 0.187 0.035 6 0.164 0.034 0.260 0.064 0.191 0.035 0.224 0.047

7 0.134 0.030 0.148 0.035 0.174 0.024 0.129 0.032 8 0.166 0.035 0.216 0.050 0.133 0.026 0.161 0.030

9 0.230 0.048 0.274 0.066 0.259 0.060 0.254 0.057 10 0.153 0.032 0.169 0.041 0.162 0.025 0.181 0.039

11 0.241 0.059 0.256 0.055 0.240 0.069 0.269 0.071 12 0.130 0.027 0.225 0.047 0.186 0.031 0.260 0.054 13 0.229 0.058 0.242 0.060 0.229 0.052 0.253 0.061

14 0.136 0.031 0.172 0.038 0.194 0.030 0.144 0.049 15 0.163 0.028 0.181 0.036 0.151 0.028 0.191 0.034

16 0.127 0.035 0.181 0.033 0.164 0.050 0.202 0.046 17 0.078 0.025 0.148 0.040 0.081 0.022 0.148 0.028

18 0.245 0.060 0.270 0.064 0.296 0.061 0.248 0.059 19 0.113 0.040 0.196 0.066 0.170 0.039 0.236 0.057 ALL 0.168 0.064 0.206 0.066 0.178 0.062 0.211 0.063

As with the pointing task, participants tended to become faster across the two blocks, both

with and without the stick. As with the pointing task study, performance was greater without

the use of the mahl-stick than with it. A repeated measures ANOVA was performed on the

Time to Complete Tasks, the Distance from the Centre of Pointing Target, and the Overall

Speed to Complete Tasks. No results of significance were detected (p > 0.05), and again no

learning effect was observed in using the mahl-stick.

Below the Index of Difficulty for the four possible line lengths are calculated, using Accot &

Zhai's formulation IDAZ. Given that the path widths were constant the entire length and

entirely straight, there was no benefit to be gained by using other formulations. The size of

the targets did not change from 24mm, only the distance between targets varied. HV1 and

HV2 are paths that existed either vertically or horizontally, one or two targets away

respectively. D1 and D2 are diagonal paths at 45º, one or targets away respectively as well.

The length L of the lines are given in millimetres.

L IDAR

HV1 135 5.625 HV2 270 11.250

D1 190.918 7.955 D2 381.838 15.910

55

4.3. Qualitative Results

In evaluating the use of a mahl-stick in performing pointing and steering tasks, qualitative

information from participants was gathered regarding their opinions on how mahl-sticks

affected performance. This data was obtained in two ways. Whilst the participants were

performing the tasks, Information of their fatigue, how they used the mahl-stick, and which

hand they held the stylus with were also recorded. This information is explored in sub-section

4.3.1. After completing both the pointing and steering tasks, participants were asked fill out a

questionnaire asking for some defining information, their past experience with a mahl-stick,

and their opinions and preference as to the performance of the mahl-stick in pointing and

steering tasks. The questionnaires are analysed in sub-section 4.3.2.

4.3.1. Observations of Participants

Participants were observed on how they used the stick to complete the pointing and steering

tasks, and how fatigued they were by completion. This information is tabled below. Fatigue is

measured in three categories: Low Fatigue, Some Fatigue, and High Fatigue. Some

participants stated they did not have issue with fatigue in performing the study, and these

people are listed under Low Fatigue. On the other hand, some participants said they had a lot

of fatigue in performing the pointing and steering tasks and are listed under High Fatigue.

Some Fatigue is a category for those in between. It's possible that some of those listed as Low

Fatigue should be listed higher due to understatement by participants.

Pointing Technique Steering Technique Fatigue

2 Pickup at start; changed to Slide Slide High Fatigue 3 Pickup Slide Some Fatigue

4 Pickup Pickup Low Fatigue 5 Slide Pivot Low Fatigue

6 Slide Slide Some Fatigue 7 Slide Slide Some Fatigue 8 Slide Slide Low Fatigue 9 Pivot Pivot Low Fatigue 10 Slide Slide Some Fatigue

11 Slide Slide High Fatigue 12 Slide Slide and Pivot Low Fatigue 13 Slide and Pivot Slide and Pivot Low Fatigue 14 Pickup and Slide Pickup and Slide High Fatigue

15 Pivot Pivot Some Fatigue 16 Pickup Ruler to start; changed to Pickup High Fatigue 17 Pivot Pivot Some Fatigue

18 Pivot Pivot Low Fatigue 19 Slide and Pivot Slide and Pivot Low Fatigue

56

Participants used the mahl-stick in different ways. Pickup refers to the motion of taking the

mahl-stick away from the wall for every task and moving it into position. Slide is when the

hand rested on the mahl-stick in an unmoving position, and the stick and hand slid across the

surface of the wall. Pivot is when the end of the mahl-stick was held in place, and the stick

was rotated and the hand slid across it to complete the tasks. Slide and Pivot refers to a fluid

motion of both sliding the mahl-stick across the surface of the wall whilst also rotating it to

affect the reach of the height. Pickup and Slide refers to the motion of picking up the mahl-

stick to get to the start of the task, and then sliding into position to complete it. One

participant started by using the mahl-stick as a ruler and not a support for the hand.

Only two participants changed the way they used the stick in the middle of either of the

studies: number 2 changed from a Pickup technique to a Slide technique during the pointing

task study, and number 16 changed from the very slow and fatiguing Ruler technique to a

Pickup technique during the steering tasks. Some participants used a different technique for

each of the studies. Number 3 changed from a Pickup technique when pointing to a Slide

technique when steering; number 5 changed from Pivot when steering to Slide when pointing;

number 12 changed from Slide and Pivot when steering to Slide when pointing.

Pickup is a slow and fatiguing way to use the mahl-stick. Of those to use the stick in this

manner, three stated they had a lot of fatigue (3, 14, 16: High Fatigue), one stated they were

getting fatigued after the first study was completed (3: Some Fatigue), and one reported

suffering no arm fatigue at all (4: Low Fatigue). The only other participant that reported

being very sore after completing the studies used the Slide technique, but stated they get very

sore writing on a whiteboard and avoided do so where possible.

57

4.3.2. Questionnaire Results

After completing both the pointing and steering tasks, participants were asked to complete a

questionnaire asking their age; gender; past experience with mahl-sticks; how easy, accurate

and fast they thought it was to complete both tasks with and without the mahl-stick; and their

preference for using the mahl-stick to complete tasks. Past experience was given the options:

I have never used one before (N)

I have some past experience (Y-)

I use one regularly (Y+)

How easy, accurate and fast they felt it was to complete the tasks was rated on a nominal

scale 1 to 5 with 1 marked as difficult, inaccurate and slow, and 5 marked as easy, accurate

and fast. The preference for using a mahl-stick for the tasks was rated on a nominal scale 1 to

5 with 1 marked with and 5 marked without.

Questionnaire Results

Age

Gen

der

Have U

sed

Mah

l-st

ick

Pri

or

Ease

-

Poin

tin

g W

ith

ou

t

Ease

-

Poin

tin

g W

ith

Ease

- S

teer

ing W

ith

ou

t

Ease

- S

teer

ing W

ith

Acc

ura

cy -

Poin

tin

g

Wit

hou

t

Acc

ura

cy -

Poin

tin

g W

ith

Acc

ura

cy -

Ste

erin

g W

ith

ou

t

Acc

ura

cy -

Ste

erin

g W

ith

Sp

eed

- P

oin

tin

g W

ith

ou

t

Sp

eed

- P

oin

tin

g W

ith

Sp

eed

- S

teer

ing W

ith

ou

t

Sp

eed

- S

teer

ing W

ith

Pref

eren

ce -

Poin

tin

g T

ask

s

Pref

eren

ce -

Ste

erin

g T

ask

s

2 29 M N 5 3 3 4 5 4 3 4 5 3 3 3 4 2 3 40 M N 5 3 5 2 5 4 5 4 5 2 4 1 5 4

4 28 F N 5 5 3 4 4 5 3 4 5 3 4 3 3 2 5 28 M N 5 4 5 4 5 5 5 5 5 3 5 3 5 4

6 24 M N 2 3 2 3 4 4 4 4 5 4 5 4 2 2 7 33 F N 5 4 5 2 4 4 4 2 5 4 4 2 5 5

8 33 M N 3 3 5 3 5 5 5 3 5 3 5 3 4 4 9 24 M N 5 5 3 2 4 4 3 3 5 3 4 3 3 3 10 30 M N 5 2 5 2 4 4 4 4 5 2 5 2 3 3

11 27 M N 4 5 4 4 4 4 5 4 5 3 5 4 4 2 12 27 M N 5 4 4 5 5 4 4 5 5 4 5 4 5 1

13 27 M N 5 4 4 5 5 5 4 5 5 4 4 4 4 2 14 40 M N 5 4 5 4 4 4 4 4 5 4 5 4 3 3

15 24 M N 5 3 3 4 5 4 3 4 5 4 3 3 4 2 16 48 F N 4 3 4 1 5 5 5 3 5 4 5 2 5 4

17 25 M N 5 2 4 2 5 5 4 4 4 2 4 2 5 5 18 26 M N 5 4 4 3 5 5 3 3 5 4 5 4 4 3 19 26 M Y- 5 5 5 5 4 4 4 4 5 4 5 4 3 1

58

Means of Survey Results

min max Mean ( ) SD (σ)

Age 24 48 29.944 6.611

Ease - Pointing Without 2 5 4.611 0.850 Ease - Pointing With 2 5 3.667 0.970

Ease - Steering Without 2 5 4.056 0.938 Ease - Steering With 1 5 3.278 1.227

Accuracy - Pointing Without 4 5 4.556 0.511

Accuracy - Pointing With 4 5 4.389 0.502 Accuracy - Steering Without 3 5 4.000 0.767

Accuracy - Steering With 2 5 3.833 0.786

Speed - Pointing Without 4 5 4.944 0.236 Speed - Pointing With 2 4 3.333 0.767

Speed - Steering Without 3 5 4.444 0.705 Speed - Steering With 1 4 3.056 0.938

Preference - Pointing 2 5 3.944 0.938

Preference - Steering 1 5 2.889 1.231

On average participants thought that the mahl-stick made pointing and steering tasks harder,

less accurate and slower. When asked to rate their preference for using the stick for pointing

tasks, participants overwhelmingly came out against it. This result was as hypothesised: for

pointing tasks alone a mahl-stick can only slow a user down and block their line of sight, and

the added stability the stick provides is not effectively utilised. On the other hand after only a

relatively short experience with a mahl-stick, participants revealed a preference for using the

mahl-stick in steering tasks even though participants thought it slower, inaccurate and more

difficult to perform tasks. Fatigue seems to play a large role in this result: the support a mahl-

stick gives to the user's hand is better utilised when drawing. Participant 8, who rated a

preference in steering tasks to not using the stick (rated at 4), offered up in comments that

there was less strain on his arm when using the stick.

59

4.4. Summary of Results

The focus of this research was to evaluate the effect that a mahl-stick has when performing

simple pointing and steering tasks in Spatial Augmented Reality applications when using a

stylus. The hypotheses for this were that the mahl-stick would hinder the performance of

pointing tasks and would improve accuracy of steering tasks at the expense of the speed they

were completed at. A repeated measures ANOVA analysis of both pointing and steering task

results did not uncover a significant result (p > 0.05) in either case.

What this research did uncover was that fatigue is a large concern in pointing and steering

tasks when drawing on vertical surfaces. Even in a relatively short trial of 30 minutes with

short breaks between blocks and the opportunity to rest half-way, arm fatigue became an

issue for many people in completing these tasks.

This research also uncovered that 30 minutes is far too short a time for people to get

comfortable in using a mahl-stick. Initial constraints of the steering study had to be relaxed

considerably so that members of the general public would be able complete the study without

a significant number of errors; however by increasing the size of both the targets and the

tunnels the overall quality of data the level of accuracy required to complete the study was

reduced considerably, which would have affected the results obtained.

I had two sub-questions in relation to the research question of this thesis:

How do novice and experienced users compare with the use of a mahl-stick in

performing pointing and steering tasks.

What learning effect is observable in using a mahl-stick.

Unfortunately I was only able to get one participant who had used a mahl-stick before. As

such there is not enough information to answer the first sub-question. The repeated measures

ANOVA analysis of both pointing and steering data did not reveal a significant learning

effect in the use of mahl-stick.

60

5. CONCLUSION

Spatial Augmented Reality (SAR) is a form of mixed reality in which computer-generated

imagery is projected directly onto real-world objects in real time, with the goal of being

interactive and to combine real and virtual objects together. As new SAR applications are

created, new input methods need to be devised to interact with them. This research was an

investigation into how a mahl-stick, a centuries-old painter's and signwriter's support for the

brush hand, could be used to perform pointing and steering tasks within a SAR system.

Pointing, or selection, tasks have become a ubiquitous selection method within computer

systems. Steering, or drawing, is a fundamental part of the creative process.

5.1. Pointing Tasks

Pointing tasks are ubiquitous in most modern computer systems. Tapping virtual buttons on a

smart phone and mouse clicking are two common pointing tasks people perform daily. A user

study was conducted in which a stylus was used to tap highlighted targets on a wall in front

the user to evaluate the performance of pointing tasks. Participants were asked to tap the

point of the stylus on the wall of highlighted targets as quickly and accurately as they could.

This was repeated twice with a mahl-stick and twice without. The time taken to reach each

target and the ultimate distance from the target was recorded.

A repeated measures ANOVA analysis was performed on the results. No significant learning

effect was discovered (p > 0.05). I hypothesised that the mahl-stick offers few advantages in

performing pointing tasks. A mahl-stick supports a painter to make precise brush strokes. A

pointing task does not require accuracy over the length of travel. As such a pointing task

would not leverage the stability a mahl-stick provides, and that the stick itself could restrict

the field of vision of the user. Participant response confirmed this; they though the mahl-stick

made pointing tasks harder, slower and less accurate, and overwhelmingly preferred not to

use a mahl-stick for such tasks.

61

5.2. Steering Tasks

Steering tasks are commonly performed in both computer systems and in the real world.

Lines are drawn extensively in creative expression and to communicate information

efficiently. A user study was conducted simultaneously with the pointing task study to

evaluate the effect that a mahl-stick has in performing simple straight-line steering tasks.

Participants were asked to draw lines as quickly and accurately as they could along a

projected path between specified projected points on a wall. The time taken to reach the end

target and the total length of the line were recorded, as well as whether or not they did so

without leaving the path.

As with the pointing task, a repeated measures ANOVA analysis was performed and no

significant learning effect was discovered (p > 0.05). I hypothesised that lines drawn with a

mahl-stick would be more accurate but slower than those drawn without one. This was not

observed in the final data. The length of time the study was run under was insufficient for

people to obtain enough practice with the mahl-stick to perform steering tasks quickly and

accurately. However participant response to the mahl-stick, even with such limited exposure,

was more positive in performing steering tasks with it than without. This confirms that a

mahl-stick is useful tool for people to perform steering tasks in SAR applications in even

simple applications.

5.3. Future Directions & Final Comments

This dissertation focused on researching the effectiveness of mahl-sticks in simple, small-

scale pointing and steering tasks. Mahl-sticks are awkward to use at first and it takes time to

get comfortable enough to use them properly. Unfortunately only one person who had used a

mahl-stick before participated in the user studies. Without adequate practitioners, the results

obtained cannot be reflective of mahl-sticks effect in performing these tasks.

Future studies of mahl-sticks should focus on training users in their use for some time to

obtain more accurate results, or run participants already familiar with their use. It takes time

to get competent using a mahl-stick, and the lack of current participant experience with them

led to insignificant results. Participants could be trained in the use of a mahl-stick by having

them draw lines, both straight and curved, with the assistance of the mahl-stick. Repeated

62

practice, perhaps two or more hours, would give participants a far greater level of

competence in the use of the stick before starting user studies of these forms. It would also

allow the accuracy required in the tasks to be increased, which would provide a more noted

and appreciable comparison between using and not using a mahl-stick.

This research was focused on short, straight-line distances between evenly spaced targets. In

the future it could be increased in scale, and made to incorporate curved lines as well. The

steering tasks tested could also be made more practical and less theoretical in future studies

by comparing how participants fare in drawing letters and simple shapes with and without a

mahl-stick would offer insight into drawing tasks.

Future studies could also analyse how the physical form of the mahl-stick affects pointing

and steering tasks. All participants used the same mahl-stick during the user studies

conducted as part of this research. It would be interesting to learn how different lengths,

thicknesses, and capping on the rested end influence how well certain tasks can be completed.

If significant differences in performance, or in peoples' preferences, were discovered it could

lead to the development of a retractable mahl-stick whose length can be adjusted to suit

specific tasks.

This research has however revealed that fatigue has an influence on task performance and it

takes only a short time before it affects users in a negative fashion. If users are to interact

with SAR systems manually for even short period of time, they need to be designed in such a

way that they can be interacted with at rest, or incorporate some form of unobtrusive support

for the user.

63

REFERENCES

Accot, J & Zhai, S 1997: ‘Beyond Fitts' Law: Models for Trajectory-Based HCI Tasks’, CHI

'97 Proceedings of the ACM SIGCHI Conference on Human Factors in Computing Systems,

pp. 295-302

Accot, J & Zhai, S 2001: ‘Scale Effects in Steering Law Tasks’, CHI '01 Proceedings of the

ACM SIGCHI Conference on Human Factors in Computing Systems, pp. 1-8

Agrawala, M, Beers, AC, McDowall, I, Fröhlich, B, Bolas, M & Hanrahan, P 1997: ‘The

Two-User Responsive Workbench: Support for Collaboration Through Individual Views of a

Shared Space’, SIGGRAPH '97 Proceedings of the 24th

Annual Conference on Computer

Graphics and Interactive Techniques, pp. 327-332

Appert, C, Chapius, O & Beaudouin-Lafon, M 2008: 'Evaluation of Pointing Performance on

Screen Edges', AVI '08 Proceedings of the Working Conference on Advanced Visual

Interfaces, pp. 119-126

Azuma, RT 1997: ‘A Survey of Augmented Reality’, Presence: Teleoperators and Virtual

Environments 6, no. 4, pp. 355-385

Azuma, RT, Baillot, Y, Behringer, R, Feiner, S, Julier, S & MacIntyre, B 2001: ‘Recent

Advances in Augmented Reality’, Computer Graphics and Applications, IEEE, vol. 21, iss.

6, pp. 34-47

Ball, R, North, C & Bowman, DA 2007: ‘Move to Improve: Promoting Physical Navigation

to Increase User Performance with Large Displays’, CHI '07 Proceedings of the SIGCHI

Conference on Human Factors in Computing Systems, pp. 191-200

Bandyopadhyay, D, Raskar, R & Fichs, H 2001: ‘Dynamic Shader Lamps: Painting on

Movable Objects’, Proceedings. IEEE and ACM International Symposium on Augmented

Reality, 2001, pp. 207-216

Beardsley, P, van Baar, J, Raskar, R & Forlines C 2005: ‘Interaction Using a Handheld

Projector’, Computer Graphics and Applications, IEEE, vol. 25, iss. 1, pp. 39-43

Bennett, E & Stevens, B 2005: ‘The Effect that Touching a Projection Augmented Model has

on Object-Presence’, Proceedings. Ninth International Conference on Information

Visualisation, 2005, pp. 790-795

Bier, EA, Stone, MC, Pier, K, Buxton, W & DeRose, TD 1993: ‘Toolglass and Magic

Lenses: The See-Through Interface’, SIGGRAPH’93 Proceedings of the 20th

Annual

Conference on Computer Graphics and Interactive Techniques, pp. 73-80

Bhatnagar, DK 1993: Position Trackers for Head Mounted Display Systems: A Survey,

University of North Carolina at Chapel Hill

64

Bimber, O, Fröhlich, B 2002: ‘Occlusion Shadows: Using Projected Light to Generate

Realistic Occlusion Effects for View-Dependent Optical See-Through Displays’, ISMAR

2002. Proceedings of the International Symposium on Mixed and Augmented Reality, 2002,

pp. 186-319

Bimber, O, Gatesy, SM, Witmer, LM, Raskar, R & Encarnação, LM 2002, ‘Merging Fossils

Specimens with Computer-Generated Information’, Computer, vol. 35, iss. 9, pp. 25-30

Bimber, O, Encarnação, LM & Schmalsteig, D 2003: ‘The Virtual Showcase as a New

Platform for Augmented Reality Digital Storytelling’, ECVE ’03 Proceedings of the

Workshop on Virtual Environments 2003, pp. 87-95

Bimber, O, Coriand, F, Kleppe, A, Bruns, E, Zollman, S & Langlotz, T 2005:

‘Superimposing Pictoral Artwork with Projected Imagery’, SIGGRAPH ’05 ACM

SIGGRAPH 2005 Courses, no. 6, pp. 16-26

Bimber, O & Raskar R 2005: Spatial Augmented Reality: Merging Real and Virtual Worlds,

A. K. Peters, Ltd. Natick, MA, USA

Bimber, O & Emmerling, A 2006: ‘Multifocal Projection: A Multiprojector Technique for

Increasing Focal Depth’, IEEE Transactions on Visualization and Computer Graphics, vol.

12, iss. 4, pp. 658-667

Brooks Jr., FP 1999: ‘What’s Real About Virtual Reality?’, IEEE Computer Graphics and

Applications, vol. 9, iss. 6, pp. 16-27

Cao, X & Balakrishnan, R 2006: ‘Interacting with Dynamically Defined Information Spaces

using a Handheld Projector and a Pen’, UIST '06 Proceedings of the 19th

Annual ACM

Symposium on User Interface Software and Technology, pp. 225-234

Cruz-Neira, C, Sandin, DJ & DeFanti, TA 1993: ‘Surround-Screen Projection-Based Virtual

Reality: The Design and Implementation of the CAVE’, SIGGRAPH '93 Proceedings of the

20th

Annual Conference on Computer Graphics and Interactive Techniques, pp. 145-132

Faconti, G & Massink, M 2007: ‘Analysis of Pointing Tasks on a Whiteboard’, Interactive

Systems. Design, Specification, and Verification Lecture Notes in Computer Science Volume

4323, 2007, pp. 185-198

Feiner, S, MacIntyre, B, Haupt, N & Solomon, E 1993: ‘Windows on the World: 2D

Windows for 3D Augmented Reality’, UIST '93 Proceedings of the 6th

Annual ACM

Symposium on User Interface Software and Technology, pp. 145-155

Fitts, PM 1954: ‘The Information Capacity of the Human Motor System in Controlling the

Amplitude of Movement’, Journal of Experimental Psychology, vol. 47, no. 6, pp. 381-391

65

FitzMaurice, GW, Ishii, H & Buxton W 1995: ‘Bricks: Laying the Foundations for Graspable

User Interfaces’, CHI '95 Proceedings of the SIGCHI Conference on Human Factors in

Computing Systems, pp. 442-449

Forlines, C, Balakrishnan, R, Beardsley, P, van Baar, J & Raskar, R 2005: ‘Zoom-and-Pick:

Facilitating Visual Zooming and Precision Pointing with Interactive Handheld Projectors’,

UIST '05 Proceedings of the 18th

Annual ACM Symposium on User Interface Software and

Technology, pp. 73-82

Forlines, C, Wigdor, D, Shen, C & Balakrishnan, R 2007: ‘Direct-Touch vs. Mouse Input for

Tabletop Displays’, CHI '07 Proceedings of the SIGCHI Conference on Human Factors in

Computing Systems, pp. 647-656

Graham, ED & MacKenzie, CL 1996: ‘Physical versus Virtual Pointing’, CHI '96

Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, pp. 292-

299

Grossberg, MD, Peri, H, Nayar, SK & Belhumeur, PN 2004: ‘Making One Object Look Like

Another: Controlling Appearance Using a Projector-Camera System’, CVPR 2004.

Proceedings of the 2004 IEEE Computer Society Conference on Computer Vision and

Pattern Recognition, 2004, vol. 1, pp. 452-459

Grossman, T & Balakrishnan, R 2005: ‘A Probabilistic Approach to Modeling Two-

Dimensional Pointing’, ACM Transactions on Computer-Human Interaction (TOCHI), vol.

12, iss. 3, pp. 435-459

Grundhöfer, A, Seeger, M, Hantsch, F & Bimber, O 2007: ‘Dynamic Adaptation of Projected

Imperceptible Codes’, ISMAR '07 Proceedings of the 2007 6th

IEEE and ACM International

Symposium on Mixed and Augmented Reality, pp. 1-10

Hinckley, K, Pausch, R, Goble, JC & Kassell, NF 1994: ‘Passive Real-World Interface Props

for Neurosurgical Visualization’, CHI '94 Proceedings of the SIGCHI Conference on Human

Factors in Computing Systems, pp. 452-458

Inami, M, Kawakami, N, Sekiguchi, D & Yanagida, Y 2000: ‘Visuo-Haptic Display Using

Head-Mounted Projector’, Virtual Reality, 2000. Proceedings. IEEE, pp. 233-240

Jota, R, Nacenta, MA, Jorge, JA, Carpendale, S & Greenberg, S 2010: ‘A Comparison of Ray

Pointing Techniques for Very Large Displays’, GI '10 Proceedings of Graphics Interface

2010, pp. 269-276

Kabbash, P, Buxton, W & Sellen, A 1994: ‘Two-Handed Input in a Compound Task’, CHI

'94 Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, pp.

417-423

66

Kato, H & Billinghurst, M 1999: ‘Marker Tracking and HMD Calibration for a Video-Based

Augmented Reality Conferencing System’, Proceedings of the 2nd

IEEE and ACM

International Workshop on Augmented Reality ’99, pp. 85-94

Kobayashi, M & Igarashi, T 2008: ‘Ninja Cursors, Using Multiple Cursors to Assist Target

Acquisition on Large Screens’, CHI ’08 Proceedings of the SIGCHI Conference on Human

Factors in Computing Systems, pp. 949-958

Kopper, R, Bowman, DA, Silva, MG & McMahan, RP 2010: ‘A Human Motor Behaviour

Model for Distal Pointing Tasks’, International Journal of Human-Computer Studies, vol. 68,

iss. 10, pp. 603-615

Kry, PJ & Pai, DK 2008: ‘Grasp Recognition and Manipulation with the Tango’, 10th

International Symposium on Experimental Robotics (ISER'06), pp. 551-559

Kurz, D, Häntsch, F, Große, M, Schiewe, A & Bimber, O 2007, ‘Laser Pointer Tracking in

Projector-Augmented Architectural Environments’, ISMAR ’07 Proceedings of the 6th

IEEE

and ACM Internationl Symposium on Mixed and Augmented Reality, pp. 19-26

Lantz, E 1996: ‘The Future of Virtual Reality: Head Mounted Displays Versus Spatially

Immersive Displays (panel)’, SIGGRAPH ’96 Proceedings of the 23rd

Annual Conference on

Computer Graphics and Interactive Techniques, pp. 485-496

Lee, J, Su, V, Ren, S & Ishii, H 2000: ‘HandSCAPE: A Vectorizing Tape Measure for On-

Site Measuring Applications’, CHI ’00 Proceedings of the SIGCHI Conference on Human

Factors in Computing Systems, pp. 137-144

Lee, JC, Dietz, PH, Maynes-Aminzade, D & Hudson, SE 2004: ‘Automatic Projector

Calibration with Embedded Light Sensors’, UIST ’04 Proceedings of the 17th

Annual ACM

Symposium on User Interface Software and Technology, pp. 123-126

Leitner, J, Haller, M, Yun, K, Woo, W, Sugimoto, M & Inami M 2008: ‘IncreTable: A Mixed

Reality Tabletop Game Experience’, ACE '08 Proceedings of the 2008 International

Conference on Advances in Computer Entertainment Technology, pp. 9-16

Low, KL, Welch, G, Lastra, A & Fuchs, H 2001: ‘Life-Sized Projector-Based Dioramas’,

VRST ’01 Proceedings of the ACM Symposium on Virtual Reality Software and Technology,

pp. 93-101

MacKenzie, IS & Buxton, W 1992: ‘Extending Fitts’ Law to Two-Dimensional Tasks’, CHI

'92 Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, pp.

219-226

MacKenzie, IS, Kauppinen, T & Silfverberg, M 2001: 'Accuracy Measures for Evaluation

Computer Pointing Devices', CHI '01 Proceeedings of the SIGCHI Conference on Human

Factors in Computing Systems, pp. 9-16

67

Majumder, A & Welch, G 2001: ‘Computer Graphics Optique: Optical Superposition of

Projected Computer Graphics’, EGVE'01 Proceedings of the 7th Eurographics Conference

on Virtual Environments & 5th Immersive Projection Technology, pp. 209-218

Marner, MR & Thomas, BH 2010: ‘Tool Virtualisation and Spatial Augmented Reality’,

Proceedings of the 20th

International Conference on Artifical Reality and Telexistence, pp.

104-109

Marner, MR, Smith, RT, Porter SR, Broecker, MM, Close, B & Thomas, BH 2011: ‘Large

Scale Spatial Augmented Reality for Design and Prototyping’, Handbook of Augmented

Reality, pp. 231-254

Marner, MR 2013, Physical-Virtual Tools for Interactive Spatial Augmented Reality,

University of South Australia

Milgram, P & Kishino, F 1994: ‘A Taxonomy of Mixed Reality Visual Displays’, IEICE

Transactions on Information Systems, E77-D, No. 12, pp. 1321-1329

Mine, MR, Brooks Jr., FP & Sequin, CH 1997: ‘Moving Objects in Space: Exploiting

Proprioception in Virtual-Environment Interaction’, SIGGRAPH '97 Proceedings of the 24th

Annual Conference on Computer Graphics and Interactive Techniques, pp. 19-26

Mistry, P & Maes, P 2009: ‘SixthSense: A Wearable Gestural Interface’, SIGGRAPH ASIA

’09 SIGGRAPH ASIA 2009 Sketches, no. 11

Mohan, A, Woo, G, Hirua, S, Smithwick, Q & Raskar R 2009: ‘Bokode: Imperceptible

Visual Tags for Cambera Based Interaction from a Distance’, ACM Transactions on

Graphics (TOG) - Proceedings of ACM SIGGRAPH 2009, vol. 28, iss. 3, no. 98

Myers, BA, Bhatnagar, R, Nichols, J, Peck. CH, Kong, D, Miller, R & Long, AC 2002:

‘Interacting at a Distance: Measuring the Performance of Laser Pointers and Other Devices’,

CHI '02 Proceedings of the SIGCHI Conference on Human Factors in Computing Systems,

pp. 33-40

Nakazato, Y, Kanbara, N & Yokoya, N 2005: ‘Localization of Wearable Users using

Invisible Retro-Reflective Markers and an IR Camera’, Proceedings of the SPIE, vol. 5664,

pp. 563-570

Nakazato, Y, Kanbara, N & Yokoya, N 2008: ‘Localization System for Large Indoor

Environments using Invisible Markers’, VRST '08 Proceedings of the 2008 ACM Symposium

on Virtual Reality Software and Technology, pp. 295-296

Nguyen, Q, Novakowski, S, Boyd, JE, Jacob, C & Hushlak, G 2006, ‘Motion Swarms: Video

Interaction for Art in Complex Environments’, MULTIMEDIA '06 Proceedings of the 14th

Annual ACM International Conference on Multimedia, pp. 461-469

68

Oda, O & Feiner, S 2009: ‘Interference Avoidance in Multi-User Hand-Held Augmented

Reality’, ISMAR 2009. Proceedings of the 8th

IEEE International Symposium on Mixed and

Augmented Reality 2009, pp. 13-22

Olsen Jr., DR & Nielsen, T 2001: ‘Laser Pointer Interaction’, CHI '01 Proceedings of the

SIGCHI Conference on Human Factors in Computing Systems, pp. 17-22

Park, H & Park J-I 2004: ‘Invisible Marker Tracking for AR’, ISMAR '04 Proceedings of the

3rd IEEE/ACM International Symposium on Mixed and Augmented Reality, pp. 272-273

Pastel, RL 2006: ‘Measuring the Difficulty of Steering Through Corners’, CHI '06

Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, pp. 1087-

1096

Piekarski, W & Thomas, BH 2002: ‘The Tinmith System: Demonstrating New Techniques

for Mobile Augmented Reality Modelling’, AUIC '02 Proceedings of the Third Australasian

Conference on User Interfaces, vol. 7, pp. 61-70

Pinhanez, C 2001: ‘The Everywhere Displays Projector: A Device to Create Ubiquitous

Graphical Interfaces’, Ubicomp 2001: Ubiquitous Computing, Lecture Notes in Computer

Science Volume 2201, 2001, pp 315-331

Raskar, R, Welch, G, Cutts, M, Lake, A, Stesin, L & Fuchs, H 1998: ‘The Office of the

Future: A Unified Approach to Image-Based Modeling and Spatially Immersive Displays’,

SIGGRAPH '98 Proceedings of the 25th Annual Conference on Computer Graphics and

Interactive Techniques, pp. 179-188

Raskar, R, Welch, G & Fuchs, H 1998: ‘Spatially Augmented Reality’, First International

Workshop on Augmented Reality, pp. 11-20

Raskar, R, Brown, MS, Yang, R, Chen, WC, Welch, G, Towles, H, Scales, B & Fuchs, H.

1999: ‘Multi-Projector Displays using Camera-Based Registration’, Proceedings of the

Conference on Visualization ‘99: Celebrating Ten Years, pp. 161-168

Raskar, R & Beardsley, P 2001: ‘A Self-Correcting Projector’, IEEE Computer Vision and

Pattern Recognition (CVPR), vol. 2, pp. 504-508

Raskar, R & Low, KL 2001: ‘Interacting with Spatially Augmented Reality’, AFRIGRAPH

'01 Proceedings of the 1st International Conference on Computer Graphics, Virtual Reality

and Visualisation, pp. 101-108

Raskar, R, Welch, G, Low, KL & Bandyopadhyay 2001: ‘Shader Lamps: Animating Real

Objects with Image-Based Illumination’, Proceedings of the 12th

Eurographics Workshop on

Rendering Techniques, pp. 89-102

69

Raskar, R, van Baar, J, Beardsley, P, Willwacher, T, Rao, S & Forlines, C 2003: ‘iLamps:

Geometrically Aware and Self-Configuring Projectors’, SIGGRAPH '03 ACM SIGGRAPH

2003 Papers, pp. 809-818

Raskar, R, Beardsley, P, van Baar, J, Wang, Y, Dietz, P, Lee, J, Leigh, D & Willwacher, T

2004: ‘RFIG Lamps: Interacting with a Self-Describing World via Photosensing Wireless

Tags and Projectors’, ACM Transactions on Graphics (TOG) - Proceedings of ACM

SIGGRAPH 2004, vol. 23, iss. 3, pp. 406-415

Raskar, R, Nii, H, deDecker, B, Hashimoto, Y, Summet, J, Moore, D, Zhao, Y, Westhues, J,

Dietz, P, Barnwell, J, Nayar, S, Inami, M, Bekaert, P, Noland, M, Branzoi, V & Bruns, E

2007: ‘Prakash: Lighting Aware Motion Capture using Photosensing Markers and

Multiplexed Illuminators’, ACM Transactions on Graphics (TOG) - Proceedings of ACM

SIGGRAPH 2007, vol. 26, iss. 3

Rasmussen, C, Toyama, K & Heger GD 1996, Tracking Objects by Color Alone, Yale

University

Schwerdtfeger, B & Klinker, G 2008: ‘Supporting Order Picking with Augmented Reality’,

ISMAR '08 Proceedings of the 7th IEEE/ACM International Symposium on Mixed and

Augmented Reality, pp. 91-94

Schwerdtfeger, B, Pustka, D, Hofhauser, A & Klinker, G 2008: ‘Using Laser Projectors for

Augmented Reality’, VRST '08 Proceedings of the 2008 ACM Symposium on Virtual Reality

Software and Technology, pp. 134-137

Seo, B-K, Lee, M-H, Park, H, Park, J-I & Kim, YS 2007, ‘Direct-Projected AR Based

Interactive User Interface for Medical Surgery’, 17th

International Conference on Artificial

Reality and Telexistence, pp.105-112

Shoemaker, G, Tang, A & Booth, KG 2007: ‘Shadow Reaching: A New Perspective on

Interaction for Large Displays’, UIST '07 Proceedings of the 20th

Annual ACM Symposium on

User Interface Software and Technology, pp. 53-56

Stevens, B, Jerrams-Smith, J, Heathcote, D & Callear, D 2002: ‘Putting the Virtual into

Reality: Assessing Object-Presence with Projection-Augmented Models’, Presence, vol. 11,

no. 1, pp. 79-92

Suganuma, A, Ogata, Y, Shimada, A, Arita, D & Taniguchi, R 2008: ‘Billiard Instruction

System for Beginners with a Projector-Camera System’, ACE ’08 Proceedings of the 2008

International Conference on Advances in Computer Entertainment Technology, pp. 3-8

Summet, J, Abowd, GD, Corso, GM & Rehg, JM 2005: ‘Virtual Rear Projection: Do

Shadows Matter?’, CHI EA ’05 CHI ’05 Extended Abstracts on Human Factors in

Computing Systems, pp. 1997-2000

70

Sutherland, IE 1965: ‘The Ultimate Display’, IFIP Congress, pp. 506-508

Sutherland, IE 1968: ‘A Head-Mounted Three Dimensional Display’, Proceedings of AFIPS

68, pp. 757-764

Uchiyama, H & Marchand, E 2011: ‘Deformable Random Dot Markers’, IEEE International

Symposium on Mixed and Augmented Reality, pp. 35-38

Uchiyama, H & Saito, H 2011, ‘Random Dot Markers’, 2011 IEEE Virtual Reality

Conference (VR), pp. 35-38

Underkoffler, J, Ullmer, B & Ishii, H 1999: ‘Emancipated Pixels: Real-World Graphics in the

Luminous Room’, SIGGRAPH '99 Proceedings of the 26th Annual Conference on Computer

Graphics and Interactive Techniques, pp. 385-392

Verlinden, J, Horváth, I & Nam, T-J 2009: ‘Recording Augmented Reality Experiences to

Capture Design Reviews’, International Journal on Interactive Design and Manufacturing

(IJIDeM), vol. 3, iss. 3, pp. 189-200

Wagner, D 2007: Handheld Augmented Reality, Graz University of Technology

Welch, G & Foxlin, E 2002: ‘Motion Tracking: No Silver Bullet, but a Respectable Arsenal’,

IEEE Computer Graphics and Applications, vol. 22, iss. 6, pp. 24-38

Wellner, PD 1993: ‘Interacting with Paper on the DigitalDesk’, Communications of the ACM

- Special Issue on Computer Augmented Environments: Back to the Real World, vol. 36, iss.

7, pp. 87-96

Witmer, BG & Singer MJ 1998: ‘Measuring Presence in Virtual Environment: A Presence

Questionnaire’, Presence: Teleoperators and Virtual Environments, vol., 7 no. 3, pp. 225-240

Yang, H & Xu, X 2010: ‘Bias Towards Regular Configuration in 2D Pointing’, CHI '10

Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, pp. 1391-

1400

Zaeh, M & Vogl, W 2006, ‘Interactive Laser-Projection for Programming Industrial Robots’,

ISMAR '06 Proceedings of the 5th IEEE and ACM International Symposium on Mixed and

Augmented Reality, pp. 125-128

Zhang, Z 2000: ‘A Flexible New Technique for Camera Calibration’, IEEE Transactions on

Pattern Analysis and Machine Intelligence, vol. 22, iss. 11, pp. 1330-1334

Zhang, X, Zha, H & Feng, W 2012: ‘Extending Fitts’ Law to Account for the Effecot of

Movement Direction on 2D Pointing’, CHI '12 Proceedings of the SIGCHI Conference on

Human Factors in Computing Systems, pp. 3185-3194

71

Zhou, X & Ren, X 2010: ‘An Investigation of Subjective Operational Biases in Steering

Tasks Evaluation’, Behaviour & Information Technology, vol. 29, no. 2, pp. 125-135

Ziola, R, Grampurohit, S, Landes, N, Fogarty, J & Harrison, B 2010: ‘Examining Interaction

with General-Purpose Object Recognition in LEGO OASIS’, 2011 IEEE Symposium on

Visual Languages and Human-Centric Computing, pp. 65-68

Zollman, S & Bimber, O 2007: ‘Imperceptible Calibration for Radiometric Compensation’,

Eurographics 2007, pp. 61-64