IMPLEMENTING AUGMENTED REALITY SANDBOX IN GEODESIGN: A FUTURE

Abstract

Geodesign method and tools are extensively used for collaborative decision making focused on different fields such as transportation, land use, and landscape and has been applied in various places around the world. Nowadays, Augmented Reality (AR), Virtual Reality (VR) and more recently AR sandbox are increasingly becoming very popular particularly as a pedagogical tool. This research aims to investigate whether an AR sandbox could enhance the understanding of people around the development of design proposals and their impacts. We explored if AR sandbox could be implemented in a collaborative geodesign workflow. We reported an experiment where people were asked to build new trails using the sandbox and how the trails they designed were integrated with a larger design. Results explore opportunities and limitations of implementing AR sandbox in a collaborative geodesign workflow based on the experiment in this paper. Our AR sandbox experiment revealed a wide range of benefits to participants in the trail planning and to the geodesign structure.

Full text

IMPLEMENTING AUGMENTED REALITY SANDBOX IN GEODESIGN: A FUTURE
FOR GEODESIGN
A. Afrooz 1*, H. Ballal 2, C. Pettit 1
1 Faculty of the Built Environment, The University of New South Wales, Sydney, Australia – (a.eslamiafrooz, c.pettit)@unsw.edu.au
2 Managing Director at Geodesignhub Pvt. Ltd.- (hrishi@geodesignhub.com)
Commission VI, IV/9
KEY WORDS: Geodesign, Augmented Reality sandbox, 3D modelling, trail design
ABSTRACT:
Geodesign method and tools are extensively used for collaborative decision making focused on different fields such as
transportation, land use, and landscape and has been applied in various places around the world. Nowadays, Augmented Reality
(AR), Virtual Reality (VR) and more recently AR sandbox are increasingly becoming very popular particularly as a pedagogical tool.
This research aims to investigate whether an AR sandbox could enhance the understanding of people around the development of
design proposals and their impacts. We explored if AR sandbox could be implemented in a collaborative geodesign workflow. We
reported an experiment where people were asked to build new trails using the sandbox and how the trails they designed were
integrated with a larger design. Results explore opportunities and limitations of implementing AR sandbox in a collaborative
geodesign workflow based on the experiment in this paper. Our AR sandbox experiment revealed a wide range of benefits to
participants in the trail planning and to the geodesign structure.
* Corresponding author
1. INTRODUCTION
1.1 Introduction
The integration of the design process with new technology has
been advocated by a group of scholars and technologists
including Bill Miller, an architect and engineer at ESRI and
Carl Steinitz, an urban designer professor at Harvard University
(ESRI 2010). The origins of geodesign dates back to 1960s with
the publication of “Design with nature” by McHarg (1969)
(Haddad 2015). Steinitz proposed many ideas and he defined
geodesign as “changing geography by design” (Steinitz 2012).
Geodesign is a “methodology” that provides a design
framework (Steinitz 2012). Steinitz (2012) defined geodesign as
“a set of concepts and methods that are derived from both
geography and other spatially oriented sciences, as well as from
several of the design professions, including architecture,
landscape architecture, urban and regional planning,...” (p.1). In
other words, geodesign is based on geographic sciences, and
interactions and negotiations between professionals and the
people of the place. It is based on data, analysis, and design
(Miller 2012).
On the one hand, there are many visualisation techniques to
support place based analysis (Pettit et al. 2012). In recent times
we are seeing a growing body of research and development in
Augmented Reality (AR), and Virtual Reality (VR) in the
context of city planning and design (Jiang et al. 2018).
On the other hand, there are many forms of thinking such as
verbal, hypothetical, statistical and so on. In science or any field
multiple forms of thinking are being used. Spatial thinking is
one form of thinking and is a collection of cognitive skills
(National Academies Press (U.S.) 2006). However, spatial
thinking – a form of human cognition which can be used in
reading urban planning and architectural blueprints (Liben
2007) - is usually challenging for people. Due to this reason
different laboratories around the world are utilizing AR sandbox
to allow students to be quickly immersed in the learning process
through a more intuitive approach. This innovative 3D
visualisation technique and real-time augmented user interface
proved to allow students to understand and create the real world
in urban planning and design (Petrasova et al. 2015) hydrology
(Petrasova et al. 2015), geoscience (Kreylos et al. 2016) and
geography (Jenkins et al. 2014) in visualising and analysing
different themes such as flooding hazards, soil erosion,
watershed development, viewshed analysis, coastal modelling
and trail planning (Petrasova et al. 2015).
This study is one of the first empirical studies that is
concentrated on the implication of the AR sandbox in geodesign
structure. Looking at an example of geodesign workshop in
Sydney, Australia (Pettit et al. 2017) this paper attempts to
bring a more intuitive approach in engaging participants in
future geodesign workshops by proposing Augmented Reality
(AR) sandbox. Geodesignhub and AR sandbox are tools that
provide support to planning and visioning processes. One of the
goals for this research was to test the effectiveness of these tools
in comprehension and the quality of interventions developed.
We are interested in the application of AR sandbox as a tool to
help people better understand and engage in place based design.
Accordingly, the focus of the paper is on the role of AR
sandbox as an interface to various components of the geodesign
process. In other words, this paper is proposing and evaluating
an AR Sandbox visualisation approach for supporting the
geodesign collaborative approach which could be used in future
geodesign workshops. The main reason for using tools like
Geodesignhub and AR Sandbox are to help the participants
develop a deeper understanding of the problems, the design
tradeoffs. These tools provide intuitive interfaces to enable
interactions. The primary objective was not to do advanced
ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume IV-4, 2018
ISPRS TC IV Mid-term Symposium “3D Spatial Information Science – The Engine of Change”, 1–5 October 2018, Delft, The Netherlands
This contribution has been peer-reviewed. The double-blind peer-review was conducted on the basis of the full paper.
https://doi.org/10.5194/isprs-annals-IV-4-5-2018 | © Authors 2018. CC BY 4.0 License.
5

spatial analysis (although it is possible to do given the digital
nature of these tools) but to invite the participants to negotiate
about the future of the place.
To investigate the aim of the study, two geodesign systems
focusing on (i) tourism and (ii) active transport are selected
from the completed Sydney Botany Bay Geodesign workshop
(Pettit et al. 2017). An experiment was designed, with
participants tasked with building new “trails or pathways” for
the fore mentioned two systems. Although trail design is a
product of expert knowledge and site surveying, spatial thinking
support through geospatial modelling can be used for this
purpose (Petrasova et al. 2015).
This paper is organized in five parts. First, it describes the
geodesign workflow and framework, and the AR sandbox.
Second, in methods section, we provide background material on
a study area from the first geodesign workshop in Australia
which was held in Sydney 2016. A case study is selected within
the previous study area and an experiment is conducted for this
area which is known as “Malabar headlands”. Participant
profiles and the process of the experiment are explained in this
section. Third, results of the online questionnaire are described.
Fourth, a discussion of the findings is presented. We propose as
to where in the geodesign framework AR sandbox integration
can be useful and supportive. The capabilities and limitations of
the AR sandbox resulted from the trail planning experiment are
explained in this section. Finally, conclusions and
recommended future research directions are outlined.
2. LITERATURE REVIEW
The geodesign framework is described in this section and later
in the paper (in Sections 4 and 5) is compared with the results
of the experiment to develop the conceptual framework of this
paper. In addition, AR sandbox and its applications in similar
projects are reviewed.
2.1 Geodesign
In 2015, the “Steinitz framework” was transformed into its
digital representation through a software that enables a digital
design workflow and it was tested in several workshops.
(Rivero et al. 2015; Ballal 2015; Nyerges et al. 2016).
Geodesignhub (Ballal n.d.) is a software platform where most of
this analytical thinking and collaboration approach takes place.
Geodesignhub is a cloud-based collaboration platform which
has been designed for carrying out projects to address decision
making in the context of complex geo-strategy problems. The
software has been used to manage sites in diverse contexts:
marine management, tourism development and so on. In this
case it was used in the context of urban design. It is often used
in the form of an interactive hands-on workshop meeting (Pettit
et al. 2017).
What makes geodesign with geodesignhub unique is the process
of creation of a collaborative design using the Steinitz
framework (Steinitz 2012). The workflow guides the
participants through a series of steps to facilitate negotiations
using software support to compare the interventions.
Geodesignhub embodies the systems-based approach to design
where the design problem is broken down in to constituent
systems or themes. The participants initially design exclusively
in different “systems” such as high-density housing, low density
housing, active transport, tourism, and so on, then synthesizes
the designs. Afterwards, they negotiate and come up with one or
a set of interventions as the best and final design options. This
collaborative process is supported by software in 2D; however,
in some cases 3D modelling of the final negotiation plan can be
prepared using JavaScript and/or CityEngine using API
connections (an example of Sydney workshop 2016).
Steinitz (2012) proposed a comprehensive framework for
geodesign. The framework asks six questions and has six
corresponding models as follow:
1. “How should the study area be described?” (Representation
models);
2. “How does the study area operate?” (Process models);
3. “Is the current study area working well?” (Evaluation
models);
4. “How might the study area be altered?” (Change models);
5. “What differences might the changes cause?” (Impact
models); and
6. “How should the study area be changed?” (Decision models)
(Steinitz 2012).
Each of the abovementioned iterations is based on a loop
diagram followed by six new questions, concepts, and graphs.
Representation model helps geodesign study to identify the
minimum required and relevant data. It also considers how
change will be visualized. Understanding the processes that are
involved in geographic change helps to identify the required
data for a geodesign study. Process model can range from direct
process models, to more complex such as temporal (“what if?”),
adaptive (“from what to what?”) and behavioural (“from
whom/where to whom/where?”) (Steinitz 2012). Geodesign
heavily relies on evaluation maps (Steinitz 2012). The concept
of evaluation models is derived from decision models and will
directly influence the change model because the design needs to
focus on the areas that need change or need to be conserved.
Evaluation models can evaluate the characteristics of the
environment qualitatively. One key challenge of the change
model is to get from present to the best possible future. Change
has four phases including vision, strategy, tactics, and actions
(Steinitz 2012). Impact model assesses the benefits and costs of
the changes quantitatively. Impact models have to be assessed
in different ways usually with a set of models such as
economics and environmental impact assessments. Finally,
decision model is where decisions are made based on the
cultural, personal, and institutional knowledge of the decision
makers.
The evaluation and change models are the two models that the
research team assume AR sandbox can play an important role
for participants to understand their designs. This will be
examined further in this paper.
2.2 Augmented Reality Sandbox
The AR sandbox was first developed by UC Davis, California
as a result of a NSF-funded project with the aim of teaching
earth science concepts (UC Davis 2016). It displays a dynamic
topographic map which composes of a box containing real sand,
a projector, and a Microsoft Kinect 3D camera which can be
connected to a computer system.
The sand is overlain by the digital projection of the contour
lines and colour elevation map. Data can be send through the
Microsoft Kinect 3D camera into either Ubuntu (system 76
2018; UC Davis 2016) or GrassGIS (NCSU GeoForAll Lab
2016) and into a software program that displays the information
ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume IV-4, 2018
ISPRS TC IV Mid-term Symposium “3D Spatial Information Science – The Engine of Change”, 1–5 October 2018, Delft, The Netherlands
This contribution has been peer-reviewed. The double-blind peer-review was conducted on the basis of the full paper.
https://doi.org/10.5194/isprs-annals-IV-4-5-2018 | © Authors 2018. CC BY 4.0 License.
6

onto the sand through the projector. The user can manipulate
the sand and simultaneously observe the real time changes of
the elevation map and the contour lines projected onto the sand.
In other words, the user shapes the real sand which is then
augmented in real time by contour lines, elevation colour maps,
and simulated water. By holding the hands under the Kinect 3D
camera, the user can add virtual water to the surface of the sand
flowing over the real surface of the sand with real-time water
simulation (Kreylos et al. 2016).
More than 150 laboratories all around the world are installing
and using AR sandbox in various fields in both education and
practice (Kreylos et al. 2016). AR sandbox can teach many
geographic concepts to users such as reading and interpreting
contour lines and topographic maps, flooding and formation of
watershed and can also be used in field trip preparation and trail
planning (Kreylos et al. 2016).
In a trail planning and sandbox study, Petrasova et al. (2015)
utilized tangible landscape (NCSU GeoForAll Lab 2016) to
calculate the optimized route between some way points. They
computed the least cost route between a selected numbers of
waypoints considered a specific slope value, construction cost,
aesthetics and view using network analysis, GrassGIS (GRASS
GIS 2018). Similar to Petrasova et al. (2015), in the design of
this study, slope degree has been calculated and some selection
criteria for waypoints have been considered in selecting the case
study such as aesthetic and environmental variables.
3. METHODS
This section describes the Botany Bay Geodesign workshop
followed by the design of the AR sandbox experiment. The
progress and purpose of the workshop have been published in
(Pettit et al 2017). This section summarises the output of the
workshop that are required for the current paper.
3.1 Botany Bay geodesign workshop
The workshop was held from 1st to 2nd December 2016 at
Sydney, Australia. A public lecture was given by Prof. Carl
Steinitz on the 30th November 2016 as a briefing for the
geodesign workshop. It included an overview of geodesign
framework with several examples from previous workshops. A
number of 30 professionals were participated the workshop.
Participants had various professional backgrounds from
different governmental and private sectors: local councils
including Randwick City Council, City of Botany Bay Council,
and Waverley Council, the greater Sydney Commission, Sydney
Water, Land and housing corporation, Transport for New South
Wales, department of planning NSW, Urban Growth, University
of New South wales, and University of Canberra, and private
companies such as Ernst and Young (EY) and Arup.
Participants were briefed of the case study (Figure 1), objectives
of the workshop and the Sydney 2050 projections.
Geodesignhub provides critical functionalities to enable
collaborative design and negotiations. The participants have to
go through three primary processes (all done together in
public): - Review existing conditions and draw ideas for
improving it using simple diagrams
- Get grouped in different teams where they pick
specific diagrams they prefer
- Compare contrast the selections form alliances and
negotiate.
Figure 1. Study area of the geodesign workshop, Sydney
December 2016; The case study of the sandbox experiment is
displayed in circle (in red)
Participants were using the Geodesignhub (Ballal n.d.) to draw
diagrams (i.e. simple polygons illustrating the location of the
project or policy) representing the proposed projects and
policies which were agreed between team members (Figure 2).
They were briefed on how to log in and use this online platform
and each team was equipped with one person with geodesign
experience.
Figure 2. An example of the projects and policies that
participants have created during the workshop using
geodesignhub (Ballal n.d.)
The workshop was run in two phases: a) scoping, data
collection, and analysis; at this stage data was collected from
relevant organization, and was assessed with the consultation of
the participated organizations; and b) implementation;
participants were involved at this stage (Pettit et al. 2017). As
the result of phase 1, nine systems were identified including:
medium density housing, high-density housing, commerce and
industry, public transport, active transport, green infrastructure,
blue infrastructure, education, and tourism. Participants were
first divided into nine groups each focusing on one system.
They were then divided into six multidisciplinary teams for
working on specific development scenarios. After evaluating
their design concepts (i.e. scenario design), participants were
presented their work and after negotiations across teams, they
came up with the final version of the scenario design (see Pettit
et al. 2017).
ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume IV-4, 2018
ISPRS TC IV Mid-term Symposium “3D Spatial Information Science – The Engine of Change”, 1–5 October 2018, Delft, The Netherlands
This contribution has been peer-reviewed. The double-blind peer-review was conducted on the basis of the full paper.
https://doi.org/10.5194/isprs-annals-IV-4-5-2018 | © Authors 2018. CC BY 4.0 License.
7

Due to the limitations of assessing the entire workshop, only
two systems of tourism and active transport were selected from
this workshop to be further examined with AR sandbox as a
trial in this paper.
3.2 Augmented Reality Sandbox experiment
3.2.1 Case study: The case study was selected from the
geodesign workshop as a site with different steep and evaluation
with potential opportunities for future tourism and active
transport system. The site was intentionally selected closer to
the coastal area for this paper to allow participants study the
erosion and other environmental sensitivity factors such as
flooding. This site is located at Sydney’s east between Malabar
and Maroubra beaches. There exist scenic coastal walkways in
the Sydney’s east. Malabar coastal walkway has been recently
opened to public. It is also known as Malabar headland national
parks. The elevation and contour lines of the case study are
shown in Figure 3.
Figure 3. Hillshade illustration of the case study for the AR
sandbox experiment. Contour line values are displayed.
3.2.2 Participants: Four participants voluntarily attended
this experiment. The corresponding author disseminated the
recruitment email to the faculty of built environment HDR
students. Four PhD students were recruited for this experiment
based on their available time, experience, knowledge, and
willingness to participate. Two participants were at each group
of tourism and active transport. Participants were PhD students
at the faculty of Built Environment, UNSW Sydney Australia
with professional expertise in either of the following fields of
study including: urban planning, urban design, architecture,
and/or landscape architecture (age range 35-44).
3.2.3 Process: The AR sandbox experiment included 2
phases: Scenario design and Sandbox (Figure 4). The
experiment took place at the City Analytics Lab (CAL), UNSW
(UNSW Built Environment 2018) in April 2018. Multi-touch
screen cruiser tables were used for the phase 1 of the
experiment and the Augmented reality (AR) sandbox was
utilized to facilitate the design of the phase 2 (Figure 5).
Figure 4. The AR sandbox geodesign experiment
Figure 5. Participants using the sandbox
AR sandbox was comprised of a box filled with kinetic sands,
3D scanner (Microsoft Kinect 3D camera Xbox360), a projector
(Optoma ML 750 LED 700 Lumens), and a laptop (System 76,
Ubuntu Linux). Kinetic sands were used for its adhesiveness
and moldability to sculpt models. 3D scanner captures changes
from distance to the sand surface. Using Ubuntu system, we
processed data using the commands originally developed by
Oliver Kreylos (Kreylos 2018a) of the University of California
– Davis open-source software available at (system 76 2018).
The software also project water flow simulation by holding the
hands under the 3D scanner or by assigning a keyboard to the
water flow simulation. For the purpose of this experiment,
GrassGIS (GRASS GIS 2018) was also used to project the
contour lines as well as the evaluation map of the tourism
system resulted from the geodesign workshop onto the sands.
Phase 1 was named scenario design (Figure 6). Participants
were briefed on geodesign process and were given the
evaluation map of the site which was resulted from the
geodesign workshop (Figure 7). Evaluation map or site
assessment maps are simple red/yellow/green maps that inform
participants where they can build and where they should be
careful (Ballal 2017).
Figure 6. The AR sandbox experiment (Phase 1- scenario
design)
Figure 7. The evaluation map of the active transport system
Source: (Ballal n.d.) geodesign workshop Sydney 2016.
*More details on evaluation map are provided in (Pettit et al.
2017)
45
40
0
35
30
25
20
15
10
ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume IV-4, 2018
ISPRS TC IV Mid-term Symposium “3D Spatial Information Science – The Engine of Change”, 1–5 October 2018, Delft, The Netherlands
This contribution has been peer-reviewed. The double-blind peer-review was conducted on the basis of the full paper.
https://doi.org/10.5194/isprs-annals-IV-4-5-2018 | © Authors 2018. CC BY 4.0 License.
8

They were divided into two teams of active transport and
tourism; each team focused on the topics related to the theme.
They were given some time to discuss, design, and negotiate
among themselves about the location of trails in relation to the
terrain, slope, scenic views, etc. to come up with one/two design
ideas of trails for the tourism and active transport systems (Task
1). They were given the existing trail and the contour lines of
the site. ArcGIS online was used at this stage and they were
using the cruiser interactive tables (Cruiser Interactive 2018) for
this exercise. They were then presented their scenario design to
the other team (Task 2). After negotiations with the other team
they ended up with a final trail design (Task 3- Figure 8). The
scenario design from phase 1 was downloaded as a shapefile
and then was exported into GrassGIS (GRASS GIS 2018) as an
input for phase 2.
Figure 8. Phase1, task 3: Participants were negotiating and
using the multitouch screen tables available at CAL (UNSW
Built Environment 2018)
In phase 2, sandbox, participants were asked to build a model of
the final scenario design from phase 1 onto the sandbox surface
displaying the trails of the current terrain. Using their hands
forming their design on the sand, participants were engaged and
interacted with each other during this experiment. The
experiment included six tasks (Figure 9). Prior to running the
experiment, participants were first briefed with the basic
geographic science. They were introduced contour lines and
topographic and elevation maps; for example, closer contour
lines represent steep slope and the wider contour lines are
spaced from each other, the gentler is the slope. The first task
was think aloud and explore the concept of elevation (National
Science Foundation n.d.). They were asked to get familiar with
the colour changes of the elevation map as they modify the sand
surface.
For the second task, participants were given the topographic
data and contour lines to build the site (Figure 10a). Digital
Elevation Model (DEM) 5 meter Grid of Australia was
downloaded from ELVIS (Australian Government (GeoScience
Australia) 2018) for extracting the contour lines of the selected
study area. Contour lines were extracted from DEM using
Spatial Analyst (ESRI 2017) and were then projected onto the
sandbox for participants to build the topographic site. Third task
was to build and transfer the design from phase 1 onto the
surface of the sand. The scenario design from phase 1, which
has been already imported into GrassGIS was displayed on the
sandbox for participants to build it (Figure 10b). Participants
were given some tools and scaled models such as trees, 3D
printed buildings and people to use in their design. Although it
was a trail design, they decided to give access to cars to reach
the Malabar headlands for disabled users. Car park and the road
are displayed on Figure 12.
Figure 9. The AR sandbox experiment (Phase 2 – Sandbox)
Figure 10. a) task 2, participants are building the topographic
site; b) task 3, participants are transferring their trail design onto
the sandbox
In the fourth task the first constraint was introduced (Figure
11a). Participants were first asked to predict that on which
landform the erosion will be stronger. Although they did not
have access to all the information related to erosion such as soil
type, and vegetation type, they were briefed that the steeper the
slope, the stronger erosion and deposition can occur because of
the speed of water which can carry more stuff in a higher speed
(National Science Foundation n.d.). Then they were asked to
design on a slope less than 5 degree (Figure 11b). Slope tool
(Spatial Analyst) was used to create a slope raster file of the
case study and was projected onto the sandbox for this task.
They were given some time to negotiate and come up with the
best design option, which in this exercise is the design with the
lowest impact on erosion on the slope less than 5 degrees.
Figure 11. a) task 4, slope map; b) task 4, slope map restricted
to 0-5 degrees
The fifth task was to test flooding. The second constraint was
introduced in this task. Virtual water was added to the map and
(a)
(b)
ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume IV-4, 2018
ISPRS TC IV Mid-term Symposium “3D Spatial Information Science – The Engine of Change”, 1–5 October 2018, Delft, The Netherlands
This contribution has been peer-reviewed. The double-blind peer-review was conducted on the basis of the full paper.
https://doi.org/10.5194/isprs-annals-IV-4-5-2018 | © Authors 2018. CC BY 4.0 License.
9

they were asked to observe where water flows and do any
required changes to reduce potential flooding-related issues
(Figure 12). At this stage, the participants added a bridge to the
design in order to avoid flooding. This is represented by a
yellow line in Figure 12. They used the terrain to identify the
location of the bridge which might not be possible with 2D
maps.
Figure 12. Task 5, virtual water was displayed on the site
Task 6 introduced a constraint of cost. They were given a
certain budget for this task and were only allowed to build 5km
of the trail (Figure13a). Scale bar was added to the map for this
task to measure the trail. Participants were given some time to
negotiate and come up with the final design (Figure 13b).
However, they did not change the design at this stage because
the cost was already in limits. This shows that there needs to be
a more aggressive cost in the future experiment, so they will be
forced to change the trail. At the end of the experiment,
participants received a link to the online questionnaire to fill.
Figure 13. a) task 6; b) Final design
4. RESULTS
This section describes the results of the online questionnaire.
We are not interested in assessing the final design in terms of
landscape architecture and/or urban planning. The questionnaire
provides information about the usability of the AR sandbox and
its performance in terms of decision-making, prioritizing design
interventions, and negotiations among team members for the
trail planning task in this paper.
4.1 Online questionnaire
An online questionnaire was designed with a total number of 17
questions for this experiment. The questionnaire composed of
five sections. The first section included general questions
regarding age range and the professional background and the
field of study of the participants. The second section was named
“sandbox usability”. Six questions were designed for this
section to query the usability of the sandbox. The rest of the
sections were “decision-making”, “prioritizing design
interventions”, and “negotiation”, respectively. At each of these
sections participants were asked three questions in accordance
with the abovementioned sections.
4.1.1 Sandbox usability: Sandbox usability questions
revealed the benefits and limitations of utilising this tool.
Participants were all able to recreate the terrain easily on the
sandbox (%100). They all rated the use of sandbox as
“somewhat easy” for trail planning (%100). In addition,
participants ranked their preference in drawing design concepts.
The first preference was designing on paper (66.6%), second
preference was using digital maps (66.6%). The AR sandbox
was ranked equally for the three preferences (33.3% for first,
second, and third priority). Respondents were “extremely
satisfied” to utilize the AR sandbox in the design stage. They
found that the AR sandbox was running very quickly and was
very practical for understanding the design. Selections of
respondents’ comments are presented below:
Respondent 1. Users can “quickly see
potential conflicts between ideas and
landform and drainage”’.
Respondents 4. “Interactivity and quick
visualisation of changes” is the main
benefit of using the AR sandbox in the
design process.
While participants rated AR sandbox as a useful tool, they
mentioned some limitations and difficulties in using the AR
sandbox such as the scale of the trails which required to be
adjusted with the scale of the terrain. Although this stage was
done using GrassGIS, because it took some time participants
mentioned it as one of the limitations of the AR sandbox.
Participants were also concerned about the accuracy of the
model which was moulded on the AR sandbox in terms of
elevation.
4.1.2 Decision making: All the four participants responded
positively to the question asking if AR sandbox helped them
understand the design (%100). Responses revealed that the AR
sandbox was “extremely useful” (%75) and “very useful” (%25)
when making decisions during the design process. Participants
were also asked to mention what other data or information they
needed to make decisions about trail design and where to put
the trail. Respondents included local ecology, water-related
data, budget, existing facilities and contours, site context, user
desires, and environmental constraints.
4.1.3 Prioritising design interventions: All participants
were able to prioritise different design interventions and ideas
using the sandbox (%100). They were “extremely satisfied”
(%75) and “very satisfied” (%25) in utilizing the AR sandbox in
prioritising design ideas during the experiment. Participants
commented that the AR sandbox helped them to understand the
site better, visualise vantage points, address some issues such as
drainage problems, and allow them to quickly negotiate and re-
design.
(a)
(b)
ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume IV-4, 2018
ISPRS TC IV Mid-term Symposium “3D Spatial Information Science – The Engine of Change”, 1–5 October 2018, Delft, The Netherlands
This contribution has been peer-reviewed. The double-blind peer-review was conducted on the basis of the full paper.
https://doi.org/10.5194/isprs-annals-IV-4-5-2018 | © Authors 2018. CC BY 4.0 License.
10

4.1.4 Negotiation: All participants were able to negotiate
their ideas among their peers using the AR sandbox (%100) and
were all “extremely satisfied” (%100) with using the AR
sandbox in providing a negotiation space between team
members. In responding to the question regarding how the AR
sandbox helped them to negotiate with their team members,
they mentioned that the AR sandbox “encouraged discussion”
(Respondent 1), “allow practical changes in short time”
(Respondent 2), enabled everyone to “touch the sandbox at the
same time” (Respondent 3), and allow them to “quick[ly] try
[different] ideas and visualise results” (Respondent 4).
5. DISCUSSION
The results of the questionnaire and the experiment itself
suggested the usability of the AR sandbox in trail planning. The
results revealed what type of data is required for such an
experiment in a larger scale. The capabilities and limitations of
the AR sandbox resulted from this experiment are summarized
in Table 1. The main demerits in the trail planning experiment
were the export functionality of the AR sandbox and matching
the scale of the final design of phase 1 onto the sandbox for
casting. These factors limited the authors to export the final
design into GIS environment for further analysis. If the export
function is added to the AR sandbox, users could assess how
close is the moulded design to the existing Digital Elevation
Models (DEM) to address the accuracy issue. In addition, the
calibration of the AR sandbox, is a time-consuming process.
One solution for this is to set up the 3D scanner and the
projector on the fixed customized table attached to the sandbox
which seems to be already utilised at some centres such as UC
Davis (Kreylos 2018b).
On the other hand, the AR sandbox was found to have many
merits in support of collaborative planning, decision-making,
communication and participant engagement. It is most effective
when the AR sandbox is being used to understand the
topography of the case study with considerable differences in
elevation rather than being used on a flat site.
In addition, the authors believe that the AR sandbox
experiments can help improving different models of the
geodesign structure including: representation, process, change,
and decision models. Table 1 shows the capabilities and
limitations of the AR sandbox resulting from the trail planning
experiment.
AR sandbox characteristics
Factors supporting trail planning
Factors restricting
trail planning
-Effective technique for moulding and
casting models
-Exporting the design
-Collaborative decision-making capability
-Scale-related issues
-User interaction and experience
-Topographic-related
accuracy of the
moulded design
-Quick and simultaneous demonstration of
design changes on the sand
-Time-consuming
calibration process
-Detecting flood prone areas
-Understanding topography
-Understanding slope
-Cost-effective design
-Prioritizing design interventions
-Better understanding the context
-Ability to project different GIS data onto
the sandbox
Table 1. factors supporting and restricting trail planning using
AR sandbox
It was found that AR sandbox is a great tool for the
visualisation of data, particularly topographic, landscape, and
watershed-related data. Therefore, it could help the
representation model and the process model of geodesign
structure in order to better understand the study area. It can also
help the change model by displaying the changes of the design
on the sandbox and examining the effect of the change on the
context. Although the user can partially envisage impacts of the
design, the AR sandbox cannot be a reliable tool for assessing
impact models because of lack of simultaneous analyses of the
site. However, it can be used for decision models where the
final decision need to be made. This assertion need to be further
examined in a geodesign workshop using the AR sandbox.
6. CONCLUSION AND FUTURE WORK
This paper describes a trail planning exercise, which is based on
Steinitz (2012) geodesign framework. An AR sandbox is used
in this paper in order to assess its implications in the geodesign
workflow for the first time. Two systems of active transport and
tourism were selected from a geodesign workshop which was
held in Sydney Australia in 2016. A smaller scale site was
selected from the previous geodesign case study boundary. An
experiment was conducted at two phases of scenario design and
sandbox with four participants. The outcome of phase 1 was a
trail with specific focus on active transport and tourism. This
design intervention was then moulded onto the sandbox and
three constraints of slope, flooding, and cost were introduced to
participants. They modified the design intervention in
accordance with the constraints mentioned above. Lastly, the
final design was displayed on the sandbox (Figure 12b).
In its current form, the AR sandbox managed to successfully
create both an educational learning environment and design
environment by offering the necessary tools for visualisation,
communication, decision making, and interaction between the
team members, as well as prerequisites for the simulation of the
site. However, the AR sandbox has the potential to be enriched
with some features and tools such as export, and scale
functionalities. These additional features could assist in the
design conceptualisation as part of a geodesign workshop.
Currently, the export function is limited to scan the sandbox
model using complex python scripts in the GrassGIS
environment (GRASS GIS 2018). Therefore, the export
function would be useful in order to provide flexibility for
further analyses on the exported model in a GIS environment.
Furthermore, the scale of the design was difficulty matched with
the sandbox. Additional import extension formats to GrassGIS
compatible with other GIS software could address this issue
such as the GeoJSON format. The incorporation of these
suggestions will lead to a more comprehensive, AR sandbox
tool which can support both educational and practical
applications. Results also show that geodesignhub and AR
sandbox can act as Planning Support (PSS) tools by facilitating
discussions around scenario planning and creating new design
interventions around planning challenges (Pettit et al. 2018).
However, this needs to be further examined in a more complex
planning challenge.
We acknowledge the limitations of the AR sandbox experiment
in this paper. Because it was a trial experiment a limited number
of participants were recruited. This could trigger a response
bias. Moreover, the phase 1 of the experiment was not
conducted during a geodesign workshop neither participants
were interacting with the geodesignhub software. Therefore, we
suggest running a full experiment during a live geodesign
ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume IV-4, 2018
ISPRS TC IV Mid-term Symposium “3D Spatial Information Science – The Engine of Change”, 1–5 October 2018, Delft, The Netherlands
This contribution has been peer-reviewed. The double-blind peer-review was conducted on the basis of the full paper.
https://doi.org/10.5194/isprs-annals-IV-4-5-2018 | © Authors 2018. CC BY 4.0 License.
11

workshop with a larger cohort of respondents, ideally exceeding
30. It is also suggested that such an exercise should be
accompanied by interviews or a focus grouped discussion. This
will be pursued in future research. Finally, future work can
compare AR sandbox with different AR and VR devices and
their implications in PSS.
ACKNOWLEDGEMENTS
We would like to acknowledge Oliver Kreylos for the open
source AR sandbox scripts, and Carmela Ticzon for her support
in providing the geodesign workshop materials.
7. REFERENCES
Australian Government (GeoScience Australia), 2018. ELVIS.
Elevation Information System. http://www.ga.gov.au/elvis/.
Ballal, H., n.d. Geodesign Hub. Dublin, Ireland: Geodesign
Hub Pvt. Ltd. https://www.geodesignhub.com/, (28 February
2018).
Ballal, H., 2015. Collaborative planning with digital design
synthesis. Doctoral thesis. University College London, London.
http://discovery.ucl.ac.uk/1471177/1/HB-Thesis-Master.pdf, (7
March 2018).
Cruiser Interactive, 2018. Informing sharing reinvented.
Australia. https://www.cruiserinteractive.com.au/, checked on
(5 March 2018).
ESRI, 2010. Changing geography by design. Selected readings
in GeoDesign. United States of America.
ESRI, 2017. ArcGIS Desktop 10.5. Redlands, CA.
GRASS GIS. Version 7.4.0, 2018. GRASS Development Team.
http://grass.osgeo.org/, (1 March 2018).
Haddad, M. A., 2015. A Framework for Geodesign: Changing
Geography by Design SteinitzCarl, 2012. Redlands, CA: ESRI
Press. 208 pp. ISBN 978-1-58948-333-0. In Journal of
Planning Education and Research 35 (2), pp. 228–230. doi:
10.1177/0739456X15581606.
Jenkins, H.S.; Grant, R.; Hopkins, D., 2014. Shifting Sands and
Turning Tides. Using 3D Visualization Technology to Shape
the Environment for Undergraduate Students: American
Geophysical Union, 12.
Jiang, L.; Masullo, M.; Maffei, L.; Meng, F.; Vorländer, M.,
2018. A demonstrator tool of web-based virtual reality for
participatory evaluation of urban sound environment. In
Landscape and Urban Planning 170, pp. 276–282. doi:
10.1016/j.landurbplan.2017.09.007.
Kreylos, O., 2018a. AR Sandbox. http://idav.ucdavis.edu/
~okreylos/ResDev/SARndbox/, (8 May 2018).
Kreylos, O., 2018b. Augmented Reality Sandbox. News. The
University of California.
https://arsandbox.ucdavis.edu/info/media/, (3 May 2018).
Kreylos, O.; Kellogg, L. H.; Reed, S.; Hsi, S.; Yikilmaz, M. B.;
Schladow, G. et al., 2016. The AR Sandbox. Augmented
Reality in Geoscience Education. In American Geophysical
Union.
http://adsabs.harvard.edu/abs/2016AGUFMED51H0843K, (1
March 2018).
Liben, L.S., 2007. Education for Spatial Thinking. In William
Damon, Richard M. Lerner (Eds.): Handbook of Child
Psychology. Hoboken, NJ, USA: John Wiley & Sons, Inc.
McHarg, I.L., 1969. Design with nature: Garden City, NY:
Natural History Press.
Miller, W.R., 2012. Introducing Geodesign. The concept.
California, USA: ESRI.
National Science Foundation, n.d. Shaping Watersheds.
Augmented reality Sandbox Facilitator’s Guide.
https://arsandbox.ucdavis.edu/wpcontent/uploads/2016/11/Shap
ing-Watersheds-AR-Sandbox-Facilitation-Guide.pdf, (1 March
2018).
NCSU GeoForAll Lab, 2016. Tangible landscape.
https://tangible-landscape.github.io/index.html, (1 March
2018).
Nyerges, T; Ballal, H; Steinitz, C; Canfield, T; Roderick, M;
Ritzman, J; Thanatemaneerat, W., 2016. Geodesign dynamics
for sustainable urban watershed development. In Sustainable
Cities and Society 25, pp. 13–24. doi:
10.1016/j.scs.2016.04.016.
Petrasova, A.; Harmon, B.; Petras, V.; Mitasova, H., 2015.
Tangible Modeling with Open Source GIS. Cham: Springer
International Publishing.
Pettit, C.; Widjaja, I.; Russo, P.; Sinnott, R.; Stimson, R.;
Tomko, M., 2012. Visualisation support for exploring urban
space and place. In ISPRS Ann. Photogramm. Remote Sens.
Spatial Inf. Sci. I-2, pp. 153–158. doi: 10.5194/isprsannals-I-2-
153-2012.
Pettit, C.; Bakelmun, A.; Lieske, S. N.; Glackin, S.; Hargroves,
K. C.; Thomson, G. et al., 2018. Planning support systems for
smart cities. In City, Culture and Society 12, pp. 13–24.
Pettit, C.J, Hawken, S, Ticzon, C., 2017. Sydney Geodesign
Workshop 2016: Developing a framework for collaborative
multi-agency scenario planning, UNSW, Sydney.
doi:10.4225/53/5a8dfd09de724.
Rivero, R.; Smith, A.; Ballal, H.; Steinitz, C., 2015. Promoting
Collaborative Geodesign in a Multidisciplinary and Multiscale
Environment. Coastal Georgia 2050, USA. In: Peer reviewed
proceedings of Digital Landscape Architecture 2015 at Anhalt
University of Applied Sciences. Berlin: Wichmann, pp. 42–58.
http://gispoint.de/fileadmin/user_upload/paper_gis_open/53755
5005.pdf, (7 March 2018).
Steinitz, C., 2012. A framework for geodesign. Changing
geography by design / Carl Steinitz. Redlands, Calf: ESRI.
System 76, 2018. Do it yourself. Augmented reality sandbox.
https://system76.com/weekend project/arsandbox, (26 February
2018).
UC Davis, 2016. 3D interactive dynamic. Powerful education
tool. https://arsandbox.ucdavis.edu/, (1 March 2018).
UNSW Built Environment, 2018. City Analytics Lab (CAL).
UNSW Sydney. https://www.be.unsw.edu.au/content/city-
analytics-lab, (12 May 2018).
ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume IV-4, 2018
ISPRS TC IV Mid-term Symposium “3D Spatial Information Science – The Engine of Change”, 1–5 October 2018, Delft, The Netherlands
This contribution has been peer-reviewed. The double-blind peer-review was conducted on the basis of the full paper.
https://doi.org/10.5194/isprs-annals-IV-4-5-2018 | © Authors 2018. CC BY 4.0 License.
12