國立臺灣大學電機資訊學院資訊工程所 碩士論文
Department of Computer Science and Information Engineering College of Electrical Engineering and Computer Science
National Taiwan University Master Thesis
行動導航之示意地圖應用
Topological Map in Mobile Navigation
李培瑄 Pei-Hsuan Li
指導教授﹕趙坤茂 博士 Advisor: Kun-Mao Chao, Ph.D.
中華民國 103 年 6 月 June 2014
Acknowledgements
Completing my Master’s Degree within 2 years with a full-time job has been challenge but also a unique life experiences. My experience at NTU has been nothing short of amazing. Carrying this out would not have been possible without the help and support of many remarkable individuals around me and I would like to extend my sincerest thanks to each and every one of them.
First and foremost, I would like to thank God who has given me the power to believe in myself and pursue my dreams. I could never have done this without the faith I have in Him.
I would like to express my deepest gratitude to my advisor, Professor Kun-Mao Chao, who has been supportive since the day I started to work on TopoNav, always providing me with invaluable input and feedback of the utmost value, carefully advising me but also encouraging me to develop my own ideas. During the most difficult times when writing this thesis, he gave me the support and freedom to develop and complete this research.
I would also like to thank my colleagues and friends from Mobile01. Especially thank Jui-Sheng Chiang, the CEO of Mobile01, for providing me with the opportunity to pursue my Master’s Degree at NTU while working at his company. Also big thanks to all the participants in the experiments of my research for spending their valuable time and effort to helping me finish it.
I am extremely grateful to my beloved parents William and Julia, my sister Claire, and my grandmother for their endless and unconditional love, support, understanding, and care throughout my life. I particularly would like to thank my partner Chris for always being there to share my happiness, sadness, and dreams, for
undertaking many unforgettable and adventurous trips around the world, and for being my eternal sunshine.
行動導航之示意地圖應用
研 究 生：李培瑄 指導教授：趙坤茂 博士
國立臺灣大學 資訊工程學系
摘要
隨著科技的創新與進步，人類得以在不同地點之間更頻繁的移動，進而提高了我
們對行動導航系統的需求。在這類型的系統中，具更高互動性的數位地圖常被作
為傳統紙製地圖的替代品。我在本論文中提出：我認為現有的行動地圖介面中，
對於指引行人從 A 點移動到 B 點的需求，尚未出現最合適的解決方案。其中主
要的困難在於行動用戶擁有多變的使用情境，且行動裝置有許多不同的限制。本
論文的目標是為行人導航設計和實作出一個更實用的行動介面，並評估其可用
性。在論文中，我將使用 geolocation apps 一詞稱呼使用地圖介面的行動應用。
為了開發出具有高可用性的 geolocation app，我依照使用者導向設計(UCD)的方
法向使用者學習，進而為使用者設計。首先，為了解此類型系統面臨的問題與現
行的解決方案，我對現有的 geolocation apps 進行全面性的研究。我在需求分析
階段透過半結構化面談(semi-structured interview)和隨身觀察(user shadowing)進
一步了解使用者需求並定義設計目標。參與半結構面談與隨身觀察的受測者代表
到陌生城市中旅行的使用族群。從這些實驗結果中，我對使用者迷失方向的原因
與他們需要的地理空間資訊有了更深入的了解。我在 iOS 上實作了行動介面的
原型 TopoNav，此軟體可直接在 Apple 的 iPhone 上運行。為了評估 TopoNav 的
可用性，我對一群受測者進行現場觀察，他們代表到陌生城市旅行的使用族群。
我同時也對 Google Maps 進行相同的測試，藉此比較觀察結果。為了評估
TopoNav 的有效性和效率，受測者被要求在測試過程中執行 4 個導航和定位任
務。我主要採用了聲音錄製和放聲思考(thinking aloud)這兩種研究方法。結果表
示，在給定的使用情境內，TopoNav 具有比 Google Maps 更好的可用性。除此之
外，我從受測者那獲得了許多寶貴的意見反饋，將在未來作為改進 TopoNav 之
用。
關鍵詞：手機、導航、找路、示意地圖、人機互動、手機介面、適地性服務
Topological Map in Mobile Navigation
Student: Pei-Hsuan Li Advisor: Kun-Mao Chao Department of Computer Science and Information Engineering
National Taiwan University Abstract
As the technology innovates and expands, the mobility of human beings increases and
the uses of a mobile navigation system are growing in its popularity. Cartographic
interfaces are often used in these systems as an alternative to traditional paper maps
with higher interactivity. In this thesis, I argue that existing mobile cartographic
interface solutions are not best suited to guide a pedestrian from A to B. The main
reasons for it are the dynamic context of use for mobile users and limitations of the
mobile devices. The aim of this thesis is to design, implement, and evaluate a more
usable mobile interface for pedestrian navigation. The term geolocation apps will be
used for the mobile applications that use cartographic interfaces in this thesis. In order
to develop a geolocation app with high usability, I followed the User-Centered Design
(UCD) approach to learn from the user and design for the user. I first performed a
thorough overview of existing geolocation apps to understand the problems and
currently available solutions. To further understand the user requirements and define
the design goals, semi-structured interviews and user shadowing observations were
conducted during the requirement analysis phase of this research. The selected
participants in these two experiments represent travelers using geolocation apps to an
unfamiliar city. From the results of these experiments, I gained a deeper
understanding of the geospatial information required by users and causes of their
disorientation. Based on the findings from requirement analysis, a prototype interface,
TopoNav, was then implemented on iOS and it runs on a standard Apple iPhone. In
order to evaluate the usability of TopoNav, a field-based observation was conducted
with representative participants of travelers to an unfamiliar city. Google Maps were
also tested for comparison purpose. To evaluate the effectiveness and efficiency of
TopoNav, the participants were asked to perform four navigation and orientation tasks
during the test. Audio record and thinking aloud were the main research methods used.
The results show that TopoNav has better usability than Google Maps in the given
context of use. In addition to this, valuable feedbacks from the participants were
collected to make further improvement of TopoNav in future.
Keyword － Mobile, navigation, wayfinding, topological map, human-computer interaction, mobile interface, location-based services
TABLE OF COTENTS
Acknowledgements ... II 摘要... IV Abstract ... VI TABLE OF COTENTS... VIII LIST OF FIGURES ... X LIST OF TABLES ... XI
Chapter 1 Introduction ... 1
1.1. Background of Mobile Navigation ... 1
1.2. Topological Map ... 7
1.3. Thesis Organization ... 10
Chapter 2 Related Works ... 11
2.1. Turn-by-Turn Navigation System ... 11
2.2. Image-Based Navigation ... 12
Chapter 3 Requirement Analysis ... 15
3.1. Fundamentals of User-Centered Design ... 16
3.2. UCD Methodologies Applied in This Research ... 22
3.3. User Profiling ... 23
3.4. Interview ... 26
3.4.1. Participants ... 28
3.4.2. Findings... 30
3.5. User Shadowing ... 32
3.5.1. Participants ... 34
3.5.2. Test Areas ... 36
3.5.3. Findings... 38
3.6. Results of the Requirement Analysis ... 38
Chapter 4 Implementation and Evaluation ... 40
4.1. Design Implications from Requirements Analysis ... 40
4.2. Implementation ... 43
4.3. Evaluation ... 48
4.3.1. Testing Methodology ... 49
4.3.2. Test Tasks and Scenarios ... 50
4.3.3. Participants ... 55
4.3.4. Test Area ... 57
4.3.5. Test Session Structure ... 58
4.3.6. Research Methods ... 59
4.3.7. Test Execution ... 60
4.3.8. Analysis of the Results... 61
Chapter 5 Discussion and Conclusion ... 79
5.1. Lesson Learned ... 79
5.2. Conclusion and Future Works ... 81
Bibliography ... 83
LIST OF FIGURES
Figure 1.1 2D digital map displays the route and the user’s current location... 5
Figure 1.2 The subway map of London Underground ... 7
Figure 3.1 Norman’s stages model of actions ... 19
Figure 3.2 The five main activities of UCD (adapted from ISO 13407 [15]) ... 21
Figure 3.3 The UCD-based plan of this research. ... 22
Figure 3.4 The application selected for this study: Google Maps displaying an area in Taipei City ... 33
Figure 3.5 The test areas ... 36
Figure 4.1 Screenshots of TopoNav ... 45
Figure 4.2 Age of smartphone users in the United States (adapted from comScore MobiLens). ... 57
Figure 4.3 Age distribution of the usability test participant... 57
Figure 4.4 The selected test area in Taipei City ... 58
Figure 4.5 Performance of the navigation sub-tasks 2f and 3d ... 64
Figure 4.6 The average number of stops each participant had during navigation ... 66
Figure 4.7 Effectiveness Chart of Task 1 ... 70
Figure 4.8 Effectiveness Chart of Task 2 ... 73
Figure 4.9 Effectiveness Chart of Task 3 ... 75
Figure 4.10 Effectiveness Chart of Task 4 ... 78
LIST OF TABLES
Table 3.1 Research methods for UCD (adapted from [17, 18, 19, 10, 20]). ... 22
Table 3.2 Examples of existing geolocation apps. ... 25
Table 3.3 Characteristics of the potential target users group ... 26
Table 3.4 Profiles of the interview participants ... 29
Table 3.5 Interview participants’ experiences in fields related to this study ... 29
Table 3.6 Profiles of the user shadowing participants ... 35
Table 3.7 User shadowing participants’ experiences in fields related to this study. .... 35
Table 4.1 Profiles of the usability test participants. ... 56
Table 4.2 The time each participant used to navigate from the starting point to the destination. ... 63
Table 4.3 The number of stops each participant had during the navigation tasks. ... 66
Table 4.4 The result for all sub-tasks in Task 1 ... 69
Table 4.5 The result for all sub-tasks in Task 2 ... 72
Table 4.6 The result for all sub-tasks in Task 3 ... 75
Table 4.7 The result for all sub-tasks in Task 4 ... 77
Chapter 1 Introduction
1.1. Background of Mobile Navigation
For thousands of years, human beings have needed to find their way to where
they needed to go, and stay oriented was a matter of life and death. One wrong turn
could lead to a bottomless cliff or a nasty death from starvation. As a result, different
techniques for wayfinding were developed and used by travelers over land and sea.
These include navigating by tracking the sun or stars’ position and celestial events as
landmarks to determine one’s position, or with the help of road signs, maps, compass,
and along with the navigating person’s cognitive effort to decide on which way to go.
As the technology innovates and expands at an exponential rate, the way people
live today has become more rapid and mobile than few decades ago. People generally
commute farther, and travel much more frequently to both familiar and unfamiliar
destinations. Regardless of the traveling distance, most activities that we do in our
daily lives are related to mobility in some way - we do activities such as work, study,
shop, eat, and sleep at different places. In order to do all these activities within the
limited time we have, a careful scheduling considering place, time, and order of the
activities is required. Location-Based Services provide solutions to this need as
information and supporting tool.
Location-Based Services (LBS) are information services that use information on
the geographical location of the mobile devices to provide various services to their
users. The term LBS generally can be applied to any application that uses the user’s
location - regardless of the user’s mobility. For example, Google Maps can be used on
either a desktop computer or a smartphone to search for places based the current
location of the system being used. The term mobile Location-Based Services (mLBS)
is preferred when referring to the LBS applications running on mobile devices.
Benefitting from the location awareness, connectivity, mobility and convenience
of mLBS, novel applications have been created to assist users for their geospatial
decision-making tasks in everyday lives. These applications are typically used to
provide information or entertainment. It can be used to find the location of an object
or a person to answer questions such as “Where is the nearest gas station?” or “Where
is my friend at?” Some other common mLBS applications include fleet management,
local advertisements, location-based games, and personalized weather.
There are many different interfaces for mLBS to communicate information
between the users and the application. In this thesis, I choose to focus on the mobile
cartographic interfaces. Cartographic interfaces are often used with other navigation
aids such as text, voice, and graphics to support effective communications in mLBS.
To distinguish with other mLBS, the term geolocation apps will be used for the
mobile applications that use cartographic interfaces in this thesis.
Geolocation apps are used as geospatial decision-making tools that support
people’s need with timely location sensitive information at anytime, anywhere. One of
the general usage scenarios of geolocation apps would be travellers arriving in an
unfamiliar city. The primary task for them often is to find out where they are and
which way to go. Although it is possible to ask someone or read a map for directions,
there is no guarantee for them to reach the desired destination without getting lost. It
is at this point that the need for geolocation apps becomes more significant.
The typical users for geolocation apps could be car drivers looking for navigation
guidance, customers looking for a business, or pedestrians searching for the most
efficient route to their destinations. While expecting geolocation apps can provide
effective solutions to these problems, there still remain many factors affecting their
feasibility in the real world practice. Some of the most important factors are the
differences in context of use and users (e.g. abilities, goals, personal preferences), the
limitations of mobile devices (e.g. small screen size, difficulty of data input,
processing power and network bandwidth), and the dynamic physical conditions (e.g.
noise, light level). These challenges must be carefully addressed when designing
geolocation apps.
Geolocation apps aim to solve the geospatial problems from their users. These
problems are often related to wayfinding. The very first step of these systems often is
to answer the orientation question “Where am I?” The concept here is simple - if one
doesn’t know where she or he is, how one can find the way that he or she wants to go.
Once obtained the user’s current location, questions such as “Where can I go?”, “How
can I get there?”, “How long does it take to get there?”, “How far away is it?”, and
“How will I know when I get there?” need to be addressed to achieve successful
navigation.
New technological advances such as mobile phones, internet, and Global
Positioning System (GPS) have transformed how we find our ways in the physical
environment [22]. Now that the GPS and internet have built into the mobile phones,
one can use it as one’s own personal wayfinding device [1]. With the position
information obtained and a digital road network model, the devices can calculate and
suggest an optimal route to the desired destination. A 2D digital street map is usually
used as the cartographic representation to display the route and the user’s current
location (see Figure 1.1). To navigate with this approach, users are required to
complete an orientation task to relate the digital map to their physical environment.
However, this can be especially difficult in crowded surroundings or urban areas
where the digital map cannot represent all the complexity and details in the 3D real
world.
Figure 1.1 2D digital map displays the route and the user’s current location.
To deal with this problem and enhance the overall navigation user experience,
companies like Google and Nokia have put their efforts to collect images along many
streets in the world. For example, Google, today’s leader and innovator in map and
navigation, sent out its 360 Degree Panoramic Camera mounted vehicles to
systematically harvest visual information [23], and the images collected can be used
in map applications to ease the burden of the users from relating artificial view on the
map to physical environment. These panoramic street-level views enable users to
virtually visit the place and explore surroundings without actually been there, thus
helping the users to recognize which road to make the turn and the destination more
easily.
Another challenge for mobile navigation is to optimize and refine the map for
use on the small screen. Traditional solutions used to presenting map information on
the desktops are not always feasible on mobile devices. Because of the limited screen
size, mobile phones can only display a relative small amount of information at once.
Simply scaling down the maps made for desktops is not practical, the result might be
difficult or even impossible to read [2, 3].
Zooming and panning are commonly used to overcome the limitation of small
screen display. However, this approach has some disadvantages. First, interacting with
the map becomes more time demanding and complicated. Second, zooming-in to get a
closer look of an area on the map results in a loss of the overview, also a large amount
of details will be omitted when the users zoom out the map. As mentioned before, the
mobile devices can only handle a limited amount of information; naively overloading
all details onto a mobile device’s screen can ruin its usability and readability. In the
case of zooming-out, the small-scale map may not include all information, and its
users may find it difficult to obtain a broader understanding of the situation.
1.2. Topological Map
Topological map - a simplified map with only vital information remains and
most unnecessary details removed - provides an interesting yet greatly unexplored
alternative approach for mobile navigation. In contrast to typical street maps, where
roads and transport links are often offered in greater details, in topological map only
important information required to successfully complete the journey are displayed.
While the relation between two points is maintained, these maps usually come with no
scales at all; also direction and distance are subject to change. A subway map or a
map with driving direction to a business is good example.
Figure 1.2 The subway map of London Underground (source: https://www.tfl.gov.uk).
The subway map of London Underground (see Figure 1.2) is one of the most
well known examples for a topological map. This map was designed by Harry Beck, a
London Underground employee, who realized that the travellers did not need to know
precisely where the physical locations of the stations are, but only the topology of the
subway lines mattered to them. He simplified the map to include only stations and
straight line segments connecting them, also altered the map’s perspective to ensure
that each line and station could be clearly seen. The resulting map immediately
became popular after its release.
Another example is the map included on a business card. Instead of laying out all
streets in neighborhood, the map usually only includes major streets, cross roads, and
landmarks around that business. With a single most important purpose, the map
provides the best way to get to the business. It helps customers to quickly and easily
find the place.
I have chosen to explore how topological map can be applied to mobile
navigation. The research question is whether it is possible to represent turn-by-turn
navigation on topological map for mobile users that are easy to read and follow
despite the challenges. To my knowledge, no research has been undertaken to study
how the use of topological map can be put in mobile map navigation scenarios, yet
these maps have already been widely used under various circumstances as mentioned
above. Topological map has the advantage that the map is generally easier and clearer
to read. Instead of focusing on the geographic details, it takes what the map attempts
to communicate at first place. I do think this might enhance the cognitive readability
and usability of the map in mobile navigation, whereas mobile devices can only
display limited amount of information at once due to their small screen size.
The proposed alternative approach is TopoNav, a prototype of a navigation
system using topological map that provides only a limited amount of navigation cues.
It allows users making local searches for points of interest (POIs), such as restaurants
and shops nearby. Once the user starts to enter a search term, the TopoNav displays a
list of all matching POIs, and this allows users to quickly browse possible travel
destinations. When the user selects a desired destination, the TopoNav provides
turn-by-turn instructions on a topological map. Here, users can see a rough outline of
the calculated route, where the instructions are represented by a set of nodes and each
node interconnects to another by a line. The node indicates where changing of
direction is needed, and the line segment is the street in between two nodes. In order
to give the route diagram a direction, different graphic icons are used to denote the
nodes at point of origin and destination. It also includes a street photo from the user’s
point of view underneath the route map.
1.3. Thesis Organization
This thesis is structured as follows. In next chapter I summarize and discuss related
works in mobile navigation done by other researchers. In Chapter 3 I describe the
requirement analysis I made for TopoNav. I followed user-centered design approach
to explain the problems I attempt to solve. The implementation and evaluation are
described in Chapter 4. In Chapter 5 I discuss the conclusion and possible future
works for this research.
Chapter 2 Related Works
2.1. Turn-by-Turn Navigation System
When people travel in an unfamiliar environment, tools such as maps and
compasses were often used to help them complete the navigation and orientation tasks.
Many studies have been done and different techniques have been proposed to find the
best solution for pedestrian navigation.
Krüger et al. [4] studied on how the users acquire route and survey knowledge.
During their test, the participants were required to walk through a route with a picture
showing the right way at every intersection. After arrived at the destination, the
participants were then required to perform a recollection task recalling the locations of
every decision point. The test results show that a turn-by-turn navigation system may
reduce the amount of survey information its users collect.
Goodman et al. [5] described a turn-by-turn pedestrian navigation system for
elders. It provides users text and audio instructions with references to landmarks.
Their test participants spent significantly shorter time to complete same navigation
tasks with the proposed navigation aids than with the map. The overall navigation
workload was significantly lower and the participants got lost less often.
Based on the results of these researches, while a turn-by-turn navigation system
has advantages, it has disadvantages too. The performance of executing navigation
tasks can be improved with such a system, but the users tend to acquire less amount of
the information about their surrounding environment. This implies that if the users
have no knowledge about the surroundings, they might easily get lost when the
system suddenly malfunctioned. The accuracy of the GPS positioning feature can
affect the performance of a navigation system [6]. Even when the GPS is working as
expected, the system might still require its user to put effort in positioning himself or
herself. The GPS signal is reflected by many materials, and can be absorbed by wood
and water. Therefore, it is only more reliable for outdoor uses. It is usually not
working indoor and not reliable in forests and urban areas with tall buildings.
2.2. Image-Based Navigation
Many researchers studied the use of the photos as navigation aids in a navigation
system. Beeharee and Steed [7] described a navigation system with geo-tagged photos.
Their navigation system comes with a map view with route line and navigation
instructions, a route tab with complete route instructions, and a viewer tab shows
photos along the route. This system was then tested with and without photos. Their
results indicate that the photos had shortened the time required to complete the
navigation task and helped the users to reorient themselves more easily. The
participants used the photos to make and confirm decisions.
Hile et al, [8] described a navigation system with online geo-tagged photos. This
system has two modes - a map mode and a landmark mode. The landmark mode
shows photos of landmarks with related navigation instructions. During the test, the
participants were required to walk a route using these two modes. The test participants
found the landmarks mode useful in intersections and they would use them again.
Chen et al. [9] described a system using route videos as navigation aids. This
system provides 360-degree panoramas videos around the pre-selected landmarks
near intersections. The system displays a map of the route with thumbnails of
landmarks and a video of the route with dot moving along the path indicating the
location of the video. In addition to this, a list of navigation instructions (written
instructions and thumbnails of landmarks) is also included. They performed a test to
compare the performance of a map with videos and a map with photos. The
participants were asked to familiarize themselves with route using either the video
map or the photo map. After finished the familiarization step, the participants were
then asked to perform a virtual drive of the same route and they were asked to make
decisions on where to turn at intersections. In general, the participants performed
better with the video maps - they needed to check for instructions less often.
These researches show that photos and videos of landmarks are useful navigation
aids [24]. This method is especially useful for decision making near the intersections
and the verification of navigation choices. It should not replace the map, but to be
provided as additional geospatial information.
Chapter 3
Requirement Analysis
In Chapter 2 several related studies for mobile navigation were inspected. By
investigating the design and development methods of these apps, the decision on the
type of methodology to use in this thesis was made. In order to deliver an effective
communication of geospatial information via geolocation apps, along with the support
of technology, the needs, desires, and limitations of the end users of such systems
must be given extensive attention during the design process. An approach
User-Centered Design (UCD) fulfills these requirements, and there are an increasing
number of geolocation apps using it in their design process. I selected UCD as the
overarching methodology for this thesis.
This chapter first provides an overview of the principles of UCD. It begins with a
discussion about usability, and then several user-centered approaches are introduced.
Following on from this is a more detailed discussion about the purpose, expectations,
and research strategies of UCD.
The second part describes how UCD applied in this research. In order to create
geolocation apps with high usability, the users for such systems must be defined first.
A requirement analysis was carried out, and the user profile for the target users that
this system focusing on was created. To gain an understanding of the user’s
requirements for a geolocation app, semi-structured interviews and field-based user
shadowing were conducted. The results from these observations are included at the
end.
3.1. Fundamentals of User-Centered Design
UCD is an approach used in computer systems design. Its purpose is to address
these fundamental questions: ‘How do I understand the user?’ and ‘How do I ensure
this understanding is reflected in my system?’ [10]. In order to create a usable and
successful final product, the design process positions the end user at the center to
ensure the user can easily use the system to meet his or her needs [11]. Vredenburg et
al. defines UCD as the practices of “the active involvement of users for a clear
understanding of user and task requirements, iterative design and evaluation, and a
multidisciplinary approach” [12]. The system does not require users to adapt their
behaviors to use; instead, it is designed to support its users’ existing behaviors.
The techniques and ideas of UCD came from the early works of Gould and
Lewis (1985) [11] and Norman and Draper (1986) [13]. At that time, Gould and
Lewis proposed and discussed three basic principles for this emerging field: (1) an
early focus on understanding users and their tasks; (2) empirical measurement of
prototype by representative users; and (3) an iterative cycle of design, test and
measure and redesign. In 1988 Norman [14] further elaborated UCD and he suggested
every good design should follow these four main principles:
1. Make it easy to determine what actions are possible at any moment.
2. Make things visible, including the conceptual model of the system, the
alternative actions, and the results of actions.
3. Make it easy to evaluate the current state of the system.
4. Follow natural mappings between intentions and the required actions;
between actions and the resulting effect; and between the information that is
visible and the interpretation of the system state.
In addition to this, Norman also suggested the following seven design rules are
essential to facilitate the designer’s tasks:
1. Use both knowledge in the world and knowledge in the head. This principle
is based on the theory of mental models. The designer has a conceptual
model about the system, and during the use the user also develops a mental
model explaining the operation of the system. For the system to succeed, the
designer’s model must match with the user’s mental model. However, the
designer does not talk directly to the user. The designer can only
communicate with the user through the "system image”. The system image
is the designer’s materialized mental model, such as the system’s
appearance, operation, or the manuals included with it. The designer must
ensure that the system image is consistent with his or her conceptual model.
2. Simplify the structure of tasks. Tasks should be simple. Avoid requiring
difficult actions like planning and problem solving in them. Consider the
limits of the user’s short-term memory (STM) and long-term memory
(LTM). On average the user is able to remember five to nine unrelated
things at a short time. Keep the task consistent and provide mental aids for
easy retrieval of information from LTM.
3. Make things visible: bridge the gulfs of execution and evaluation. Consider
Norman’s stages model of actions in Figure 3.1. The left side of the figure
(intention to act, sequence of actions, and execution of actions) is the
execution part. The user interface should provide information for the user to
decide which actions he or she should undertake. The right side of the figure
(perceiving the state of the world, interpreting the perception, and
evaluation of the perception) is the evaluation part. The user interface
should provide feed to show the user what can be done and what the results
are.
Figure 3.1 Norman’s stages model of actions.
4. Get the mappings right. A natural mapping that leads to immediate
understanding should be used. The representation of the control’s
functionality should be made with careful considerations of cultural
standards and physical analogies.
5. Exploit the power of constraints, both natural and artificial. Design the
system to make the user feel that there is only one action possible or logical
to do.
6. Design for error. Assume that the user will make errors. The system should
be designed to allow the users to recover from any possible error made.
7. When all else fails, standardize. Create a universal standard for things
cannot be explained in any logical or culturally determined way.
In 1999 an international standard ISO 13407 was release, entitled
‘Human-Centered Design Processes for Interactive Systems’, providing a
standardized description of UCD [15]. ISO 13407 provides guidance for achieving a
high level of usability through UCD actions and strategies. This standard describes an
iterative cycle of development composed of five main activities of UCD. Four of
these activities are performed iteratively until the outcome meets the initially set
objectives. Every UCD project should carry out these activities from the beginning.
These activities are listed below and presented in Figure 3.2:
1. Plan the human centered process.
2. Understand and specify the context of use.
3. Specify the user and organizational requirements.
4. Produce design solutions.
5. Evaluate designs against requirements.
Figure 3.2 The five main activities of UCD (adapted from ISO 13407 [15]).
ISO 13407 aims to provide general guidelines for UCD, the detailed description
about the research methods and techniques required to conduct these activities is not
included [16]. In fact, there are many research methods and techniques available for
UCD, with Table 3.1 presenting a selection of them, grouped by the four main
activities described above. The detailed discussion about each of these methods is not
within the scope of this thesis, instead, only the details of those been selected and
considered for this research will be included in below and following chapters.
Table 3.1 Research methods for UCD (adapted from [17, 18, 19, 21, 20]).
3.2. UCD Methodologies Applied in This Research
The main objective of this thesis is to design and evaluate a cartographic UI for
geolocation apps that supports the geospatial activities of a pedestrian user in
unfamiliar surroundings. In order to achieve high usability in such a system, UCD
methods should be carefully applied. As the first step, requirement analysis was
conducted to determine the user's needs, characteristics, context of use and
preferences. Methods like user profiling, interviewing existing users, and user
shadowing were used. Regarding traditional methods such as interviewing or
surveying can fail to gain deep understanding of the use context; a field-based user
shadowing was applied additionally to users during their actual trip in unfamiliar
environments.
The UCD-based plan of this research is presented in Figure 3.3. Following this
plan, the rest of this Chapter describes the Requirement Analysis stage, including user
profiling (Section 3.3), interview (Section 3.4), and field-based user shadowing
(Section 3.5).
Figure 3.3 The UCD-based plan of this research.
3.3. User Profiling
Understanding the target users of a system is critical in the UCD approach, and a
careful selection of a representative user group for the system to focus on is required.
Geolocation apps have a wide range of users in the general population. A survey was
conducted by GlobalWebIndex to determine the most used smartphone app in the
world. A list of smartphone apps were selected and presented to the survey
participants, and they were asked to answer the question “Which of the following
mobile applications have you used in the past month?” Not surprisingly, Google Maps,
a typical geolocation app, is the most popular mobile apps in the world, with over
54% smartphone owners using at least once during the month of August 2013 [25].
Since it would be infeasible to include the entire population of the potential users, the
target user population is limited to a smaller and manageable size which only focusing
on certain types of geolocation apps. This is used as an initial description of the users
to create the user profile.
To identify a meaningful geolocation apps area to concentrate on, I conducted a
review of related researches and commercial implementations. The result shows that
the majority of both the research projects and the commercial geolocation apps
targeted on tourism and travel type applications. As tourism information is commonly
geospatial, there is a large scope for exploring usable representations of it. A selection
of these commercial travel-based geolocation apps are presented in Table 3.2. With all
these factors, this thesis will focus on the area ‘travel and navigation’ of the
geolocation apps.
Table 3.2 Examples of existing geolocation apps.
User profiling defines the use contexts, characteristics and preferences of the
potential end users who are going to use the cartographic interface. In addition to the
general profile of the users of interest in this research given in Chapter 1, a more
detailed description of the characteristics of the intended users is shown in Table 3.3.
Table 3.3 Characteristics of the potential target users group.
3.4. Interview
To further understand the user requirements and define the design goals,
semi-structured interviews were conducted. The interviews aim to assess how a user
plan and conduct their trip in an unfamiliar environment. A typical scenario can be a
traveller arriving by train in an unfamiliar city - as he or she exits the underground
railway station, it may not be easy for such a traveller to immediately know where he
or she is and which way to go. The orientation question of “where am I?” must be
addressed first before resolving all other geospatial problems such as how to reach a
desired destination from the current location.
Based on the objective of this thesis and the related works described in previous
What kinds of geospatial information are required by the travelers when
they first arrive in an unfamiliar area?
In which way do users use the mobile map to navigate?
Do users find street photos and voice directions helpful?
What information on the mobile maps and the mental maps of users are
important for navigation and wayfinding?
Do users find it difficult to relate the mobile maps with the real world
surroundings?
How often and when do users feel themselves lost orientation and what is
the reason for that?
Do users find landmarks helpful?
Do users have problems with the zooming and panning techniques in
mobile maps?
The objective of the interview is to obtain a more clear understanding about the
user needs for pedestrian navigation and wayfinding. The findings from these
interviews were later used in the design of the prototype.
3.4.1 Participants
As presented in the previously created user profile (see Table 3.3), the interview
participants should represent a particular group of geolocation app users who visit an
unfamiliar city. 10 participants are recruited to participate in this study. 5 of the
participants are female and 5 are male. Few of them are foreigners. The average age is
32.4, with a minimum of 20 and maximum of 50. All participants owned touch-screen
smartphones and had used a geolocation app like Google Maps before. However, they
may have different levels of experience and knowledge with mobile devices, digital
maps, cartography, and navigation and wayfinding techniques. Table 3.4 and Table 3.5
present the profiles of these participants and their experiences with different fields
related to this study.
Table 3.4 Profiles of the interview participants.
Table 3.5 Interview participants’ experiences in fields related to this study.
3.4.2 Findings
I discovered common user behaviors from the interviews. In general, the
majority of participants plan their trips using a computer, tablet, or guidebook at home
before they travel to an unfamiliar city, and the rest of the participants only do this if
the preparation time allows. Most of them use digital maps on the screen of a
computer or tablet, and some use the printed paper map on the guidebook. They use
the map to find POIs and the routes from the public transportation stations to their
hotels. In order to learn about all the best POIs to visit, they search on the web, find
information on travel websites like TripAdvisor or Lonely Planet, or ask others who
have already been to the city for recommendations. To prepare for the actual
navigation, they ask friends who have visited the city about important landmarks that
can help them to stay oriented. They also look for the street photos of the POIs for the
reason that it could help them to recognize the places more easily.
Upon their arrival to the unfamiliar city, some participants use paper maps to do
the navigation, some use the navigation apps on their mobile phones. Four of the
participants (P1, P4, P5, P9) find using the paper maps difficult, they sometimes still
get lost in the unfamiliar environment. Most participants find the mobile navigation
system more helpful than paper maps. They only use the paper maps when they have
no access to a mobile navigation system.
In the case of using a mobile navigation system, the participants use the mobile
phones to search for the desired destination on the go. They often use mobile devices
as a tool to figure out the next destination. For example, at restaurant, they search for
good coffee shops to go after the meal. If they are at the hotel, they commonly look
for the navigation guidance on a computer or tablet before heading out. The
destination often described as landmarks, places, or POIs. Once decided on a
destination, three of the participants start the turn-by-turn navigation immediately; the
other seven of the participants preview the navigation instructions and the route on
map before they start. Street images and landmarks are important at the decision
points along the calculated route, such as the final destination and the turning points
where the moving direction changes. If panoramic street images are provided, they
rotate the street images from different viewing angles to look for business logos and
obvious landmarks. They find the street images toward their point of view more
useful and meaningful than street images from other viewing angles.
The participants find the details available on the street map much less important;
they normally concentrate more on the highlighted route that represents the directions
for turn-by-turn navigation. They look for street names, landmarks, and the
intersections before the next turning point to make sure that they are aware of the turn
beforehand. This inspired me to remove unnecessary map details in the design of
TopoNav.
Sometimes the participants still get lost, even with the help of maps and mobile
navigation system. When this happens, if using a paper map, six of them try to find
their way by using the street name information or landmarks nearby to relocate
themselves on the map; if using a mobile navigation system, the system commonly
would recalculate the route to destination and show its user how to get back to the
correct path.
3.5. User Shadowing
In order to complement the findings from the interview and better understand the
context of use, I additionally shadowed users during their actual trips. As presented in
Table 3.2, there are already many geolocation apps available in the market.
Considering the equipment available in this requirement analysis, only those
applications compatible with iOS 7.1 and Android 4.4 were investigated.
For this study, an existing geolocation app was selected for the participants to
perform real world navigation tasks in the context of a traveller to an unfamiliar city.
The application was selected based on the following features:
Covers the study area (Taipei City)
Availabilities to the participants and researchers
Turn-by-turn navigation
Voice directions
Street images
Zooming and panning functions
Google Maps (Figure 3.4) was selected for that it supports all the features listed
above and is the most popular mobile application currently available.
Figure 3.4 The application selected for this study: Google Maps displaying an area in Taipei City.
3.5.1. Participants
Another group of participants were recruited for this study; they also belong to
the group of geolocation app users representing travelers visiting an unfamiliar city.
The participants were observed in an area where he or she was not familiar with, and
supported with the information from the selected geolocation app, Google Maps.
I recruited another 8 participants (4 female) ranging in age 22 to 36, with an
average age of 27.6 years old. All participants have used the selected geolocation app,
Google Maps, before, and they all have been to Taipei City more than once. Each
participant was given the freedom to select the test area where he or she has never
visited before. They all owned touch-screen smartphones. All rated themselves with a
high experience for mobile devices. They may have different levels of experience and
knowledge in digital maps, cartography, navigation, and wayfinding. Table 3.6 and
Table 3.7 present the profiles of these participants and their experiences with different
fields related to this study.
Table 3.6 Profiles of the user shadowing participants.
Table 3.7 User shadowing participants’ experiences in fields related to this study.
3.5.2. Test Areas
The observations took place in Taipei City. Taipei City is the capital of Taiwan. It
has lots visitor attractions to offer and is visited by many people for different purposes
such as business or leisure. Most of the visitors go there by public transportations. For
all these reasons, it is very well suited for this study.
Different test areas were selected by each individual participant. The participant
was asked to select an area that is new to him or her around any Taipei Metro (MRT)
station. These test areas are presented in Figure 3.5.
Figure 3.5 The test areas.
Participants are given a task to navigate to a random POI nearby using Google
Maps on their mobile phones. Following the principle of shadowing technique, I tried
to be unobtrusive while they performing the task. Findings were recorded after each
observation session from the observer’s memory.
Regarding the diversity of contexts in the real world, the observation sessions
were conducted under following predefined conditions to ensure the reliability of the
results:
Only during the daytime (from 8AM to 5PM)
Under benign and generally pleasant weather conditions (sunny or limited
low-level cloud cover, light winds and good visibility)
Without any unusual disturbing source such as road construction and
demonstrations
One participant was observed during each observation session. I met up with the
participant at Taipei Main MRT station and travelled with him or her by train to the
selected test area. During the train ride, a briefing about this study was given, a
mobile device (an iPhone) with Google Maps installed was provided, and the
participant was told the test would start immediately after he or she arrived the
destination station.
3.5.3. Findings
All participants start to make route selection immediately after they decided on a
destination. It does not seem to be important about which route to take; they appear to
choose the first suggested route. The digital street map highlights the chosen route and
displays only critical and major street names. Seven of the participants pan and zoom
the map to get a complete preview of the planned turn-by-turn navigation; one starts
to navigate directly. On digital map, some street names are omitted to avoid labels
overlapping each other. The participants look for hidden street names along the route.
In the case of walking, I discovered that most of the time mobile devices are left in the
pocket or hold in hand, the participants only stop to read the map when they feel
themselves lost orientation. None turns on the voice directions.
3.6. Results of the Requirement Analysis
This section summarizes the findings from the semi-structured interviews and the
user shadowing observations to answer the questions presented in Section 3.4.
Only current route matters - all participants in the interview find most details
on the traditional street map unimportant. During the shadowing observation,
seven of the participants pan and zoom the map to see the street names along the
calculated route. These results indicate that current route has an overwhelming
importance over other parts on the street map.
Identifying the destination and turning points is necessary - in the interview,
all participants report street images and landmarks helpful at the destination and
turning points. However, when I shadowed travelers, Street View was not used.
The reason might be that there is no Street View provided for turn-by-turn
navigation on the current mobile version of Google Maps. The participants are
discussing their experience from desktop Street View in the interview.
Fragmented attention to mobile devices while walking - I observed that the
voice directions are not used, and people often just holding the mobile device in
hand or keeping it in the pocket during the trip. They only interact with the
mobile device when they feet lost. From both the interview and the shadowing
observation, I learned that most participants often preview the route before they
start to navigate. It shows that route preview is a usable and effective navigation
tool for travelers.
Chapter 4
Implementation and Evaluation
Based on the user requirements collected, I found that generally the travelers
consider the calculated route significantly more important, look for an intuitive way to
identify the decision points such as destination and turning points, and only pay
fragmented attention to the mobile device while walking. I argue that existing
solutions such as digital street maps and turn-by-turn navigation systems are not the
best way to guide a pedestrian from A to B. To provide an alternative approach to this
problem, I propose TopoNav: a pedestrian navigation system on a mobile device that
uses a topological map.
4.1. Design Implications from Requirements Analysis
With the requirements analysis, I observed that travelers look for an easy way to
learn critical information about the calculated route. Instead of providing all the
unneeded details on a street map, the following describes the properties that a
pedestrian navigation system on mobile devices should follow:
Focus on the calculated route - I found that the majority of the travelers ignore
most details on the street map before and during their trip. To obtain a quick
overview of the calculated route, travelers typically look for navigation cues such
as street names and landmarks along the route. Although the calculated route is
highlighted in a typical navigation system, I believe more details on the map can
be omitted to keep users away from distractions. The system should try to focus
on providing only the information required by travelers in the first place, and a
topological map might be well suited for this purpose.
Identify the destination and turning points - from the observation, identifying
the destination and turning points are critical for any travelers who wish to
navigate to a destination. In an unfamiliar environment, travellers need to be
notified in advance when approaching a turning point or the destination along the
route. They typically look for street names, landmarks, and street images around
these locations. Therefore, the system should provide the possibility for its users
to quickly and conveniently identify these decision points.
Represent information in a clear and simple way - due to the limited display
size, modern mobile devices cannot display all the required information at once.
The observed travelers often need to repeatedly pan and zoom the map to get a
complete overview of the calculated route. Additionally, it requires travelers to
change their focus away from the activities in the real world to interact with their
mobile phones. This can be dangerous if the user interface is cluttered with
information and demands the user’s full attention. In order to provide a higher
usability while users being mobile, I assume the system should provide a user
interface that remains simple and requires minimum interactions.
No constant attention to the mobile screen is required - in the navigation
scenarios, the observed participants are constantly on the go. Their main focus is
not always on the mobile devices and they are often busy at doing something else.
While walking, they rarely check the map and never use the voice directions on
the mobile devices. Considering their short attention spans, the system should
not require the users to frequently to interact with.
Reduce cognitive burden - to navigate with a 2D map, it requires travelers to
relate the information provided by the map to their physical environment. The
travelers do not have to worry about the relationship between the map and the
surroundings with voice directions. However, I found that no participants use it
in navigation regardless of it being provided during the observation. The reason
might be that it requires users to be actively attentive to the system, and
sometimes the users may not be prepared for directions (e.g. chatting with
friends). The system should provide any kind of navigation aids that helps to
reduce the cognitive load from users.
4.2. Implementation
The TopoNav prototype is implemented on iOS and runs on a standard Apple
iPhone 5 (iOS version 7.1). With a 13 GHz dual core A6 processor, the device
provides smooth operations over existing map applications. It has built in GPS
receiver to determine the latitude and longitude coordinates of one’s current position.
Also a digital compass has been integrated to the device to obtain the direction that
one is facing, which can be useful when used with a map application. A web-based
solution is adopted which uses Google Map API to obtain geographic data for
TopoNav. Conceptually, this application can be ported to other platforms which have
built in GPS receivers and digital compasses.
With the user requirements collected I take the next step. The goal of TopoNav is
to investigate the usage of topological map in mobile navigation. I implemented
TopoNav along with following design concepts.
Quick access to the desired destination - in order to help travelers find the
destination they look for as soon as possible, I try to make the selection of the
destination in TopoNav intuitive. The destination can be searched via text-entry,
similar to traveler’s previous experience with existing map applications such as
Google Maps or Nokia Maps.
There are two rationales behind this design decision. First, the observed travelers
all have experience with existing map applications. Users often create mental
models before they start use software and these mental models come from their
previous experience with similar software. These mental models are later
referred by the users to predict what they should do with the software. Second,
instead of general areas, observed travelers often select specific business or
address as their destination. Therefore, I assume that travelers would prefer to
have text-entry search in the navigation system.
The screenshots in Figure 4.1 show the implementation of this design concept.
There is a textbox in the view. While typing the name of the business, the address
of the place, or the category that the user is looking for, a list of options matching
the text entered displays below the textbox. Tapping one of the options on the list
will select the POI as the destination, and then the system will launch route
preview from the user’s current location to the selected POI.
Figure 4.1 Screenshots of TopoNav.
Simplify map to focus on the calculated route - in order to make the TopoNav
easy to use on mobile devices, I eliminate as many unnecessary information on
the map as possible. Instead of using a traditional 2D street map, I choose to
represent the calculated route on a topological map.
From the requirement studies, the observed travelers often only looked for
information along the calculated route to destination. If travelers would like to
know how to get to the destination, they only require a general overview along
the route as guidance. It is not necessary to precisely represent every detail in the
area; instead, providing a rough outline of the calculated route in the map is
mandatory. In addition to this, simplifying the map can help to overcome the
limitations of the small screen on mobile devices. The resulting map should be
easier and clearer to read.
The user interface of the navigation view (Figure 4.1) is implemented following
this concept. Once the user selected a destination, the system calculates the route
and displays it on the map. Only the route to destination is drawn on the map,
other details, such as streets in the area, are omitted. By default, the map is at the
zoom level that the complete route can be displayed at once on the mobile device.
Based on the positioning methods available on the device, such as GPS, a blue
dot is displayed on the map to indicate the user’s current location. A red pin is
displayed on the map to indicate the destination.
The route is composited by multiple line segments. Each line segment represents
a street or road along the route. Naively placing every street name label along the
route might be problematic - labels might overlap with each other excessively,
and the resulting map is difficult or even impossible to read. To overcome this
label placement problem, labels are placed in order of their importance. Major
roads and the coming up street in next step are assigned with higher priorities. If
needed, a street name label is rotated and resized before placing beside its
corresponding line segment.
Identify turning points and destination for orientation - to provide orientation
support, keeping the users aware of the turning points and destination is critical
in a turn-by-turn navigation system. Therefore, the TopoNav prototype is
designed to help users effortlessly identify these special locations along the
route.
In order to reduce the cognitive load from users, I choose to use navigation aids
such as photographs and written instructions. On the top of the screen, there are
the navigation instructions that describe what the user should do at next turning
point (for example, turn right onto Robson Street). A street image towards the
traveler’s point of view at the next turning point is displayed underneath the
topological map. By tapping the toolbar at the bottom of the screen, the system
displays a list view of all written directions and street images, which lets the user
browse through the list for a quick summary of the route and faster switching
between the steps. In addition to this, landmarks near the turning points and
destination are displayed on the map.
Before the actual navigation starts, users can choose to preview the calculated
route. By default, the starting point is selected and the corresponding street
image and navigation instructions are displayed on the screen. As shown in
figure 4.2, gestures like swipe left and swipe right are used to select previous
step and next step on the route. As the selected step changes, the street image and
navigation instructions on screen change too. In order to differentiate the selected
line segment from the rest of lines along the route, the selected line segment is
highlighted in black color, and the rest of the route is in green color.
To minimize the downloading load required during the actual navigation, the
system pre-fetches all the street images in background after a destination has
been selected. The rationale behind this is the system’s user experience and
performance in the actual navigation can benefits from preloading images,
speeding up image load time.
4.3. Evaluation
Once the development for TopoNav prototype finished, a field-based evaluation
of its design by real world users was conducted. The usability of the prototype was
investigated to ensure that the user requirements had been met. The aim of these
usability tests was to gather user feedbacks for the cartographic interface and
determine where in the implementation need to be improved and refined. Based on the
results obtained from these tests, the usability of prototype for the target user group
can be optimized. This section describes the usability testing of TopoNav, starting
with a selection of testing methodology and the execution of these evaluation
techniques (Section 4.3.1), and then presents an analysis of the findings (Section
4.3.8).
4.3.1. Testing Methodology
There are different UCD techniques (empirical or inspection) available for
conducting a usability test [19]. Empirical methodology provides more user
involvement; the data was collected through the observations of target users
interacting with the prototype to achieve certain appointed tasks. Conversely, the
inspection method is generally performed with minimal user involvement; it only
relies on usability experts, developers, or designers to identify usability problems.
Since the objective of this thesis is to design a highly usable cartographic interface for
the target users, empirical methodology was considered to be more suitable. Only the
empirical methodology was used in this study.
Field-based observation was used to investigate the possible usability problems
of the prototype under real world conditions. As it is important to understand the
participants’ thoughts such as their intentions and expectations while they performing
the specified tasks, the technique think-aloud protocol was applied.
The purpose of test should be explained in detail to the participants before it
starts. The briefing must be made clear to them that it is the usability of the prototype
to be evaluated, not their abilities or capabilities. It is important to encourage the
participants to provide any positive or negative feedbacks about the prototype.
4.3.2. Test Tasks and Scenarios
A set of tasks was developed for the participants to perform during the evaluation
sessions. These tasks were formulated based on the general scenario of navigation
from a starting point to an unfamiliar destination. It is important for the tasks to
represent as much as possible of the uses under real world context and include the
most important parts of the user interface [21]. Furthermore, the tasks should be
structured in a way allowing the participants to compare and evaluate different user
interfaces, which means the usability of TopoNav is compared against another
geolocation app. Google Maps is selected as the geolocation app for comparison - as
TopoNav uses Google Maps API for its source of geographical data, and they both
share common characteristics such as turn-by-turn navigation
Following are the tasks developed for this evaluation:
Task 1 - Selection of Destination
Scenario 1
You are visiting an unfamiliar city for the first time. You arrived at its central railway
station, and just walked out from one of its exits. You were given the mobile interface
