See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/266654899

World of workout: a contextual mobile RPG to

encourage long term fitness
Article November 2013

CITATIONS

READS

41

9 authors, including:
Daniel Goldberg

Tracy Anne Hammond

Texas A&M University

Texas A&M University

15 PUBLICATIONS 60 CITATIONS

142 PUBLICATIONS 1,025 CITATIONS

SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate,

letting you access and read them immediately.

SEE PROFILE

Available from: Daniel Goldberg

Retrieved on: 26 July 2016

World of Workout: A Contextual Mobile RPG

to Encourage Long Term Fitness
Joey Bartley, Jonathon Forsyth, Prachi Pendse, Garrett Brown, Paul Hagseth, Da Xin
Department of Computer Science, Texas A&M University
{bart1739,jonathonforsyth,prachirp,gbrown09,pehagseth,daxin}@neo.tamu.edu
ABSTRACT

guidelines. The specific types of exercises that

integrate with these games are also limited to a
few basic categories of motions [14]. A strong
link between in-game progress and real-world
progress needs to be enhanced in the exergaming
market [1]. One-dimensional tracking systems
have been designed in several types of exergames
to track distance ran or steps taken, but these do
not provide a complete picture of physical
activity, which is crucial.

In today's digital world many individuals spend

their day in front of a computer or mobile phone
for entertainment when they have free time.
Individuals enjoy a more sedentary lifestyle from
advances in technology [2]. This is one of the
leading factors contributing to a decrease in
fitness level for the major part of the populations
in developed countries. We therefore want to
design a mobile role-playing game (RPG) where
the in game character evolves based on the
exercises the user performs in reality. This
design can motivate and persuade a potentially
large demographic of users to engage in physical
activity for an extended period of time through
the enjoyment of a fascinating game.

In this paper, we aim to provide a complete

physical activity system with continual
motivation through our exergaming model,
developed into an application called World of
Workout. This game provides a large selection of
exercises using sensors widely available in
smartphones, allowing physical activity recording
on a wide selection of devices. We also plan to
provide several channels of feedback for both
long and short-term drive, which will encourage
users to exercise regularly. A strong tie between a
users workout routine and gameplay makes the
incentive of exergaming more direct.

Author Keywords

Exergaming; Motivation; Fitness

INTRODUCTION

In the past decade mobile phone gaming has

drastically increased with smart phones allowing
the inclusion of sensors and interfaces between
the device and the user. These new mobile
interfaces aim to tie games with actions in the
physical world, encouraging users to avoid a
sedentary lifestyle. The combination of mobile
games with environmental interaction attempts to
improve physical well-being by motivating
exercise and significantly reducing the perception
of fatigue. This application field that attempts to
combine exercising and gaming have been
dubbed as exergaming. Several clear examples
of this concept include Nintendos Wii Fit,
Konamis Dance Dance Revolution, and
Microsofts Kinect console.

By having correlation between game play and the

exercises performed by the user, we encourage
progress in multiple aspects of a proper fitness
routine while keeping the user entertained. This
is an improvement on the previous single-focus
models that, for example, track only
cardiovascular statistics. We consulted health
care professionals during the development
process to verify the quality of exercise and to
make sure the recorded activities are appropriate.
Finally, we implement methods such as triggers,
randomization, and accelerated progress for
continued motivation.

Most exergaming participants play the game for a

short period of time due to lack of continuing
motivation [2].
The novelty of todays
exergaming applications either wears off quickly
or fails to provide the user with proper fitness

RELATED WORK

Many applications have been released on the

smart phone market to encourage exercise

-1-

activities but either focus solely on short or longterm motivation. For full benefit to the user, the
application should utilize both forms to
encourage the user to become and stay fit. The
key is to find a balance using a myriad of
motivation techniques in an integrated system to
captivate the user.

a web page, and sensors. The logic level is how

the application works and is divided into the
exercising and gaming engines respectively with
their added components. The last level is the data
level that stores all information and interacts with
the program through HTTP and a server.

Computer-Based Motivation

When computers were invented, the idea of using

them for the purpose of behavior modification
was considered science fiction [7]. The inability
of a static-state system to measure the users
activity patterns played a major role in supporting
this paradigm.
Existing Applications

ZombiesRun! [15] is a mobile application (app)

designed to add entertainment value to running or
jogging. The game works by playing the sounds
of a zombie apocalypse to the user as he or she
runs. It encourages the user to run faster and
farther through a scenario of critical danger. This
approach lacks a long-term motivation
component because the App cannot adapt for
long-term effect. This means that the runner can
become quickly bored of the once-novel zombie
environment. Another famous App that suffers
from this is Nike Run, which has pre-recorded
statements from famous athletes to encourage the
user.

Figure 1 System architecture representation with game and

exercise parts highlighted
Users Perspective of the Application

From the users perspective the App is divided

into three separate components. The webpage
acts solely as an interface to the database where
the user can check his or her statistics and
information. This is a convenient way to modify
the user profile and allows the individual to
review his or her data on a monitor instead of a
small phone screen. This way, the user can have a
larger display of his or her progress easily
accessible via a web interface to update his or her
current workout regime.

Walk2Build [4] is another App on the market that

encourages fitness but unlike ZombiesRun! [15],
this App lacks immediate motivational feedback.
The game records the number of steps taken by
the user and converts this number to currency in a
Sim-style world, building a city. The more the
user walks or runs, the bigger the city gets,
showing the user the results of his or her efforts.
However, getting to a large city takes a long
time, and users can fall short due to lack of
immediate rewards [5].

The second component is the game itself. It is

built using the libraries and structures of
AndEngine, which is an open-source Android
game platform. The game play entails the user
exploring and trying to survive in a postapocalyptic world, fending off monsters and other
survivors. This is done through a unique avatar
controlled by the user, which becomes stronger
based on the real world exercises the user
performs.

IMPLEMENTATION
System Architecture

Our system architecture divides the project into

three main levels as shown in Figure 1. The user
level is how the individual actually interacts with
the system, which is done through mobile phones,

The flip-side of the game is the exercise engine.

It contains a robust user interface that allows the

-2-

user to navigate and choose from a variety of

exercises to perform. The engine connects to the
server and is able to access and modify each
users exercise data in the database so that new
physical activities can be made available to the
rest of the application. In a similar manner, the
app can access the GPS and accelerometer in the
users smartphone to record the exercise.

on the exponentially increasing part of the curve,

so the user can see improvement quicker in the
game than in real life and will be encouraged to
keep working out. This technique was designed
to persuade the user to keep using the app past the
initial 2-5 week period of use. This way, people
can see the improvements in this timeframe, keep
their motivation, and press on.

When the user wants to exercise, he or she will

either browse and select a particular activity or let
the engine pick a random activity. The engine
will use the appropriate sensor to track the
number and intensity of the chosen exercise.
Once the activity is completed, the user will be
shown his or her results and the database will be
updated giving the user credit for exercises
performed.

The second advantage of using a logistic function

is an imposed bound past which additional
exercise does not improve the users in-game
character statistics. We refer to this attribute of
our app as a fatigue factor, discussed below.
A final feature of the app is a random-reward
function. After a given length of time, the game
recommends a particular exercise for the user to
perform. This helps remind the individual to
work out as well as expanding his or her
knowledge of what activities benefit progress in
the game. This trigger gives a reward by
unlocking bonus statistics for the avatar to
compel the user to stay active.

Interactions within the Platform

Gameplay

The game itself is designed to be a never-ending

survival story that is based on the style of
gameplay from previous successful games such
as World of Warcraft and Diablo. These games
have shown to retain users for extended periods
of time so we took some of their concepts and
developed them into our game. We created
several zones where characters with certain
statistics can thrive to better accommodate the
wide variety of users we plan on reaching. Some
individuals will inevitably build stronger or faster
avatars than others, and our system creates a
positive incentive for everyone by rewarding
working out in general and a variety of
approaches. Random rewards can also boost
characters so they can enjoy new realms they
were not able to previously. The end goal of the
app is to encourage the user to improve. Hence,
the game map itself mirrors increasing difficulty
and rewards for the users in game characteristics
as they become faster, stronger, and able to
endure longer.

Figure 2 In-game avatar attributes

The exercise and game engines themselves

interact with each other through three statistics:
speed, strength, and stamina. Each of these values
increases when the user performs corresponding
types of exercises. For instance, running long
distances increases stamina while sprinting
increases speed. The values that are recorded
through the phone sensors go through a logic
algorithm located on the server that determines
new statistics for the character. When the game
starts, it reads these statistics and improves the
users avatar accordingly. This direct correlation
allows users to see how each exercise helps their
personal fitness by noting what attributes are
increased.
Our logic algorithm is based on a logistic
function. This function has two attributes. The
first is accelerated improvement when the user is
-3-

mile. This also attempts to discourage users from

cheating the system. If a user tries to trick the
system by simply moving his or her foot while
sitting down to simulate running, the App will
eventually give no credit for this. The last
advantage of fatigue correlates with the
accelerometer and the ability to see when certain
activity thresholds are reached. As a person
works out, he or she gets tired and their actions
are not as accurate. For example, the user may
not reach a 90-degree angle between biceps and
forearms when doing the last pushup in a
workout. We try to communicate to the user
whenever form decomposes in a way that makes
sense and describes how to improve.

Figure 3 Screen shot of example world

Technical Considerations

To help make sure the application runs smoothly,

we attempted to isolate each of its components to
reduce background consumption of resources.
Both game and exercise engines read and update
from the database only at startup and at closing.
The game gets the values of strength, stamina,
and speed, as well as relevant game data such as
status and current bonuses then proceeds to run
without further communication until the user
decides to quit or switch modes. In this case, it
updates the database with any changes. The
exercise component has a logic system included
within the engine that updates the database with
improvements recorded at the end of each
workout. The logic rules are adjusted based on
input from our consultation with the Texas A&M
kinesiology department. This is critical for our
system since a large quantity of health apps on
the market today have no oversight from health
care professionals. In this way we have accounted
for the apps accuracy and users safety.

RESULTS

We conducted a user study on ten students at

Texas A&M who are currently working on their
undergraduate degrees. A user background form
was used to collect demographic information and
to help baseline fitness levels for our sample
users. About half of these individuals work out
more than once a week, while the rest were more
sporadic. A few had tested fitness applications
before that represent parts of the exergaming
market. Examples of these applications include
GPS Run Tracker, World of Fitness, and My
Fitness Pal.
In terms of in-game character statistics, users
rated stamina as the most important and speed as
least important to their perspectives of fitness.
They also rated a series of 15 questions on a scale
of one to five. The two highest rated questions,
both with ratings of 4.8 out of 5, were ease of use
of the GPS exercise tracking and the variety of
activities that the application supports. The
average ratings of each feature we tested for are
represented in table 1.

Fatigue factor

Our application utilizes a novel approach to a

balances workout through a fatigue factor. It is a
system of diminishing returns that encourages the
users of the app to correctly exercise in a safe
manner. This is used to make sure the user does
not overwork themselves. The app gives less
credit for the same exercises if the user is at
unsafe levels of performing those exercises. For
example, running 20 miles everyday non-stop
without letting the body rest is an unsafe exercise
and the user will see little to no improvement in
avatar statistics for the 21st mile than for the 5th
-4-

Table 1-User Study Results

Feature

Average
Rating

Accuracy of movement
tracking

4.67

Ease of GPS Use

4.83

Likeliness of GPS re-use

4.39

Jumping jack measurement

accuracy

4.13

Push-up measurement
accuracy

3.50

Cruches measurement
accuracy

4.00

Ease of Accelerometer
Measurement

3.83

Variety of Activities

4.83

Recognition of Character
Improvement

4.61

Game Enjoyment

3.89

Likelihood of Future
Gameplay

3.72

Recognized Correlation of
Exercise to Game

4.06

Likelihood of Future Exercise

4.06

Motivation to Exercise

3.94

firmware is not outdated and correcting this has

proven nearly impossible. We are currently in the
process of moving to a new server, but we have
managed to run successful tests, including
creating accounts and databases. We need a final
server before we can finish this part of the
project. To test, we will have users create an
account and comment on the process.
Our current website is a shell and is waiting for
the final data server to import all the fields it will
display. We plan on having users navigate the
site after it is connected to the database to gauge
the websites usability and ease of use as well as
user impressions of the UI we have designed.

Figure 4 Creating an account on the database and login with

website

Figure 5 shell website

Exercise engine

Rating Compared to Similar

Apps

Our exercise engine has shown a lot of promise in

recording activities. Though the database is still
not working, we have completed several tests
with the sensors. We have successfully tracked a
distance activity with the GPS. This logs the
distance traveled and saves a set of points to keep
the route using the Google API (See Figure 6). In
the study, we want users to simply move around
campus and see if they are happy with the
tracking capabilities.

4.25

Data Server and Website

The data server we have set up is currently

inefficient. The school firewall is blocks the
ports so we cannot communicate with it outside
the TAMU network. We were given a workaround to open the port, but the Cisco server
-5-

The accelerometer has shown promise when

recording data and we have been able to quantify
the numbers of exercises completed. We can also
see weak activities in a set or ones that have a
low-recorded threshold.
Being able to
differentiate particular activities has proven
difficult due to the variable sizes of users. The
following figures show some recorded data from
activities. We plan on having users perform basic
activities such as jumping jacks and pushups to
see how accurate the recorded data is as our test
of this portion of the app.

motivation to play and see results can then join

with balanced workout routines to significantly
benefit the users. Since the conclusion of the
study, we refactored sections of the game source
code to improve extensibility. From there, we
began to increase the number of enemies.
Exercise data currently displays as character
statistics live in the game, as shown below.

Showing the current progress allows for the

effective testing of different motivation systems.
Expanding the game in general leads toward our
goal of generating a higher level of user interest.
This direct display currently gives us an edge in
the development of reward calculators and
character-enemy interaction.

Figure 6 Distance demo

Pattern Recognition

After collecting accelerometer data from

activities we have found two key patterns. We
have designated these patterns W and M patterns
respectively. The W pattern results from an
activity that has a spiked motion or quick linear
motions. Weight presses are examples of a
spiking motion activity where the user quickly lift
weights straight up and down. The following
figure shows the complete accelerometer data
followed by a clean W pattern for recognition.

Figure 7 Z-axis change for various activities

Figure 8 data from pushups

Game play

Our current game engine is incomplete

DISCUSSION

As a consequence of our user study, we have

identified draw of the gameplay as the most
important aspect of this project. The effectiveness
of gamified workout applications directly ties to
the interest generated by playing the game. This
-6-

jacks. This activity has a spike motion as the user

is rapidly moving in linear paths if the user holds
the phone in his or her hand. The phone however
travels in an arc, which means we should expect
an M pattern influence. The following figure
represents an actual result from jumping jacks.

Figure 9 Spiked activity data

Figure 13 Jumping Jack data

As we hypothesized, there is an evident W pattern

in the Y and Z axes. However looking at the Xaxis shows an M pattern intertwined with the W
pattern since the activity follows an arc path. Our
ability to classify these patterns allows for easier
addition of new activities and analysis of their
results since the output usually follows a
predetermined graphical outline.

Figure 10 W pattern in Y direction

The second pattern is the M pattern that comes

from circular activities where the phone follows a
circular path. Lunges register as a circular
activity since the phone travels in an oval circuit.
The following figures show a simple circle
pattern of the phone to make it easy to visualize.

The most obvious patterns extracted from varying

exercises have given us the ability to implement
simple classification algorithms directly on the
phone. These can already identify the difference
between crunches and jumping jacks, for
example. Moreover, this activity identification
works in tandem with the live character statistics
to deliver an instant bonus to the player. As we
have noted, an exercise like jumping jacks can
not only be identified, but also improves the
corresponding aspect of the characters in-game
abilities.

Figure 11 Circular activity data

CONCLUSION

This project aims to improve upon existing

exergaming mobile applications by developing an
addictive role-playing game where a players ingame character evolves in response to exercise in
real life. Expansion of the concept of exergaming
in our research involves the combination of shortand long-term motivation. We see the
achievement of this through the instant
gratification of gameplay and the cumulative
increase in avatar abilities.

Figure 12 M pattern in Y direction

By identifying these particular patterns, we can

classify our activities into subsections for
analysis. In so doing, we will help increase the
accuracy of our application and expand the
variety of exercises that the user can perform. To
illustrate this point we will analyze jumping

-7-

The results of both our study and previous work

in mobile gaming show not only that this market
is competitive, but also that it has an edge on
console games in the realm of fitness
applications. This is due to a low price of entry
for users and the ubiquity of smartphone
hardware that supports sophisticated sensor
integration like GPS, accelerometer, electronic
compass, and hi-resolution sound and imaging.
Data from our investigation also shows how
insights pulled from accelerometer readings
provide a high level of detail.

Engineering at Texas A&M for funds and

equipment they provided.
REFERENCES

[1] Goh Kim Nee; Bakar, M.S.B.A., "Androidbased exercise application," Computer &
Information Science (ICCIS), 2012
International Conference on , vol.2, no.,
pp.1105,1109, 12-14 June 2012
[2]Svendsen, G.B.; Soholt, Y.; Munch-Ellingsen,
A.; Gammon, D.; Schurmann, A., "The
Importance of Being Useful and Fun:
Factors Influencing Intention to Use a
Mobile System Motivating for Physical
Activity," System Sciences, 2009. HICSS
'09.
42nd
Hawaii
International
Conference on , vol., no., pp.1,10, 5-8 Jan.
2009

Incorporating this information into fitness

applications can benefit the mobile exergaming
market. From the experience of our project, we
believe on-board sensors will put mobile devices
on top of the fitness application market. Our
study shows that the combination of multiple
input channels from various types of sensors
allows improved accuracy for recognizing input
from physical activity.

[3]Fogg:2009:BMP:1541948.1541999,A
behavior
model
for
persuasive
design,
Proceedings
of
the
4th
International Conference on Persuasive
Technology, Persuasive '09,2009

Overall, we found that a direct connection

between exercise and in-game progress creates an
effective motivational basis for exergaming.
Although users pointed to several areas for
improvement in our gaming and tracking systems,
their response to the incentive method gives us
confidence that mobile fitness applications will
benefit from shifting to a reflective representation
of progress. Through our test application, World
of Workout, we have found reason to augment
exergaming with reward and feedback concepts
from persuasive technology, real time responses
on a mobile platform, fatigue considerations from
sports medicine, and an addictive gaming context
in a way that keeps users engaged in long-term
fitness.

[4] Iain Hamilton, Gennaro Imperatore, Mark D.

Dunlop, David Rowe, and Allan Hewitt.
2012. Walk2Build: a GPS game for
mobile
exergaming
with
city
visualization. In Proceedings of the 14th
international conference on Human-computer
interaction with mobile devices and
services companion (MobileHCI '12).
ACM, New York, NY, USA, 17-22.
[5] Levent Grg, Abraham G. Campbell, Kealan
McCusker, Mauro Dragone, Michael J.
O'Grady, Noel E. O'Connor, and Gregory
M. P. O'Hare. 2010. FreeGaming: mobile,
collaborative, adaptive and augmented
exergaming. In Proceedings of the 8th
International Conference on Advances in
Mobile Computing and Multimedia
(MoMM '10). ACM, New York, NY,
USA

ACKNOWLEDGMENTS

This application was developed as a senior design

project at Texas A&M University. We would like
to thank Dr. Tracy Hammond, Paul Taele, Manoj
Prasad, Dustin Joubert, Dr. Dan Goldberg, and
our fellow senior design classmates for giving us
helpful feedback. We would also like to thank the
Department
of Computer Science and

-8-

[6] Mathew J. Gregoski, Martina Mueller, Alexey

Vertegel, Aleksey Shaporev, Brenda B.
Jackson, Ronja M. Frenzel, Sara M.
Sprehn, and Frank A. Treiber. 2012.
Development and validation of a
smartphone
heart
rate
acquisition
application for health promotion and
wellness telehealth applications. Int. J.
Telemedicine Appl. 2012, Article 1
(January 2012)

[10]

BJ Fogg. 2009. Creating persuasive

technologies: an eight-step design process.
In Proceedings of the 4th International
Conference on Persuasive Technology
(Persuasive '09). ACM, New York, NY,
USA

[11] Pavol Bielik, Michal Tomlein, Peter Krtky,

tefan Mitrk, Michal Barla, Mria
Bielikov. Move2Play: an innovative
approach to encouraging people to be
more physically active. In IHI '12
Proceedings of the 2nd ACM SIGHIT
International
Health
Informatics
Symposium.

[7] Helena Tobiasson, Anders Hedman, and

Yngve Sundblad. 2012. Design space and
opportunities for physical movement
participation in everyday life. In
Proceedings of the 24th Australian
Computer-Human Interaction Conference
(OzCHI '12), Vivienne Farrell, Graham
Farrell, Caslon Chua, Weidong Huang,
Raj Vasa, and Clinton Woodward (Eds.).
ACM, New York, NY, USA

[12]Shlomo Berkovsky, Jill Freyne, Mac

Coombe,
Dipak Bhandari, Nilufar Baghaei.
Physical activity motivating games: you
can play, mate! In IHI
'12 Proceedings of the 2nd ACM SIGHIT
International
Health
Informatics
Symposium.
[13] Kevin G. Stanley, Ian Livingston, Alan
Bandurka, Robert Kapiszka, Regan L.
Mandryk. PiNiZoRo: a GPS-based
exercise game for families. Futureplay '10
Proceedings
of
the
International
Academic Conference on the Future of
Game Design and Technology.

[8] Pascal Landry, Narcs Pars, Joseph Minsky,

and Roc Pars. 2012. Participatory design
for exertion interfaces for children. In
Proceedings of the 11th International
Conference on Interaction Design and
Children (IDC '12). ACM, New York,
NY, USA
[9] Kevin G. Stanley, David Pinelle, Alan
Bandurka, David McDine, and Regan L.
Mandryk. 2008. Integrating cumulative
context into computer games. In
Proceedings of the 2008 Conference on
Future Play: Research, Play, Share
(Future Play '08). ACM, New York, NY,
USA

[14] Anthony Whitehead, Hannah Johnston,

Nicole Nixon, Jo Welch. Exergame
effectiveness: what the numbers can tell
us. Proceeding Sandbox 10, Proceedings
of the 5th ACM SIGGRAPH Symposium
on Video Games
[15] Zombies, Run! 2011. Homepage.
https://www.zombiesrungame.com.

-9-