An Oculus Rift based Exergame to Improve Awareness in Disabled

People

Manuela Chessa

, Gabriele Balocchi

, Michela Busi

, Antonio Novellino

and Fabio Solari

Dept. of Informatics, Bioengineering, Robotics, and Systems Engineering,

University of Genoa, Viale Causa 13, Genoa, Italy

ETT SpA, Via Sestri 37, 16154, Genoa, Italy

Keywords:

Virtual Reality, Head-mounted Display, Walk Tracking, Active Video Games, Natural Human-computer-

interaction.

Abstract:

In this paper we present an exergame, based on the Oculus Rift head-mounted-display, with the aim of im-

proving spatial awareness in young people with cognitive deﬁcits. The scope is to create a virtual environment

that should be immersive, and should allow a natural human-computer interaction, without creating discom-

fort to the users. The exergame is simple, since the aim of the present work is not to create a photo-realistic

scenario, but a familiar environment in which to play and exercise cognitive abilities. To measure and track the

movements of the users’ legs, in order to simulate the walking in the environment in a safe way, an additional

sensor, the Playstation Move, has been embedded into the system. Finally, the system has been tested with

some disabled subjects, who conﬁrmed the usability of the exergame and a general positive feeling with such

an immersive virtual reality.

1 INTRODUCTION

In this paper, we present an exergame designed for

the Oculus Rift head-mounted-display (HMD), devel-

oped to improve spatial and cognitive awareness for

disabled people.

Exergames, or active video games, are video

games with interfaces that require active involvement

and the exertion of physical force by participants.

These exergames are designed to track body motion

and body reactions, providing both fun and exercise

for players. Numerous video game console compa-

nies have designed exergaming interfaces that have

become more and more popular over the last years.

In the last years, many researchers addressed the

problem of developing human-computer-interfaces,

often based on Virtual Reality (VR), with the aim of

obtaining immersive environments that allow a natu-

ral interaction with them. A great effort is devoted

to the graphic quality of such environments, in or-

der to achieve high levels of realism, especially when

such systems are used for training purposes (Kwon

et al., 2013), for surgery (Chan et al., 2013), or sim-

ply for entertainment. The natural interaction with

such systems has been recently considered, see for

example (Solari et al., 2013). In that paper, the au-

thors developed a stereoscopic rendering technique

that solves the misperception issues, which affect a

user free to move in front of a display, thus allowing

him/her to interact in a natural way with the virtual

objects. Other authors analyzed the issues of mis-

perception that affect HMDs (Sharples et al., 2008;

Ukai and Howarth, 2008). In particular, a fundamen-

tal problem that must be considered when creating

VR immersive environments, and that represents one

of the main issues of such kind of systems, are ad-

verse symptoms that may arise from VR use. In the

real world, when a person moves, e.g he/she changes

the position of their eyes or head, the projections of

the 3D real world immediately shift on the retinas,

and at the same time the vestibular system indicates

the movement of the head. Due to hardware and soft-

ware limitations, in HMD VR systems there is an un-

avoidable delay between a users movements and the

updating of the virtual rendered scene. If this delay

is excessive, the sensory information from the users

visual and vestibular systems might be conﬂicting,

and this can result in symptoms such as nausea, stom-

ach awareness, dizziness, and headache. Finally, an-

other important aspect to be considered is immersiv-

ity, i.e. the feeling of presence when acting in a VR

environment through head-mounted-displays (Water-

worth et al., 2010).

In this work, we do not address the problem of

770

Chessa, M., Balocchi, G., Busi, M., Novellino, A. and Solari, F.

An Oculus Rift based Exergame to Improve Awareness in Disabled People.

DOI: 10.5220/0005852607700777

In Proceedings of the 11th Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2016) - Volume 4: VISAPP, pages 770-777

ISBN: 978-989-758-175-5

graphical realism, such the exergame we present has

been intentionally kept very simple from the graphics

point of view. On the contrary, we aim to obtain a

system where the users (that are disabled people) can

act in natural and comfortable way in order to maxi-

mize the healthy effects provided by the proposed ex-

ergame.

2 STATE OF THE ART

As consumer exergaming programs have evolved, nu-

merous academics have researched the effectiveness

of these types of games on rehabilitation or exercise

(Tanaka et al., 2012). Most of the studies were con-

ducted using the Wiimote and a new peripheral, the

Wii Balance Board, and covered a lot of different sub-

jects: (i) Brain function rehabilitation (Deutsch et al.,

2008; Hsu et al., 2011; Joo et al., 2010); (ii) Isomet-

ric muscle strengthening (Sohnsmeyer et al., 2010);

(iii) Energy expenditure (Hurkmans et al., 2010); (iv)

Exercising for elderly (Wollersheim et al., 2010); (v)

Balance training (Deutsch et al., 2009; Kliem and

Wiemeyer, 2010).

In contrast, there have been a number of studies

identifying limited effects of exergaming. For exam-

ple, regarding the potential of Wii Bowling for reha-

bilitation in patients with upper extremity dysfunction

(Hsu et al., 2011), it was found that the only signiﬁ-

cant ﬁnding was a measure of enjoyment of activity

when compared to a standard exercise group. Sim-

ilarly, the study that applied the Wii games to exer-

cise and the elderly (Wollersheim et al., 2010) showed

no signiﬁcant increase in physical activity due to ex-

ergaming. Research in the medical ﬁeld has been

done for the Oculus Rift as well. A study on a pa-

tient affected by severe burn wounds on a large area of

his body (Hoffman et al., 2014) has provided the ﬁrst

evidence that entering an immersive virtual environ-

ment using the Rift can elicit a strong illusion of pres-

ence and reduce pain during the Virtual Reality expe-

rience. Other interfaces that could be used in exercise-

based games have yet to be fully explored. The use of

heart rate is being examined (Parker et al., 2011) and

galvanic skin response has been studied, as means to

communicate emotional responses (biofeedback) to a

computer. More complex biological measures, such

as electrocardiogram signals or oxygen consumption

devices, could certainly have applications here, but re-

liable devices are too expensive for home use.

3 MATERIAL AND METHODS

3.1 Hardware

The proposed exergame makes use of two external

hardware devices: the Oculus Rift and the PlaySta-

tion Move.

Oculus Rift. The Oculus Rift

was released by

Oculus VR for developers with various hardware re-

visions over the span of one year. The device used

for this project is the Developer Kit 2 (DK2). The

Oculus Rift DK2 uses an OLED panel for each eye,

each having a resolution of 960 × 1080 pixels. These

panels have a refresh rate of 90 HZ and globally re-

fresh, rather than scanning out in lines. It uses high

quality lenses to allow for a wide ﬁeld of view. The

separation of the lenses is adjustable by a dial on the

bottom of the device, in order to accommodate a wide

range of interpupillary distances. Headphones are in-

tegrated, and they provide real time spatialized binau-

ral audio. The Oculus Rift has full 6 degree of free-

dom rotational and positional tracking. This tracking

is precise, low-latency, and sub-millimeter accurate.

PS Move. The Sony PlayStation Move

is actu-

ally composed of two devices: the Move Eye and

the Motion Controller, or wand. The Move Eye

is an RGB camera (640 × 480 pixels @ 60 fps /

320 × 240 pixels @ 120 fps) with directive micro-

phones, and it’s utilized to detect an illuminating

sphere attached to the wand in order to track the con-

troller in a three-dimensional space, calculating the

distance/depth based on the size of the sphere on each

frame. For this project the Eye was not needed, thus

only the wand was connected via bluetooth to the sys-

tem. The Motion Controller contains a three-axis ac-

celerometer, a three-axis gyro sensor and a geomag-

netic sensor. The accelerometer and the gyro sen-

sor are used to track rotation in overall motion and

can be used for dead reckoning (in cases when the

camera tracking is insufﬁcient). To correct cumula-

tive errors on these sensors, the geomagnetic sensor is

used for calibrating the wand’s orientation against the

Earth’s magnetic ﬁeld. Consequently, the sensor fu-

sion method makes it possible to recognize the wand’s

position and orientation robustly and accurately.

3.2 Software

The entire project was developed with Unity in C#,

https://www.oculus.com

https://www.playstation.com

An Oculus Rift based Exergame to Improve Awareness in Disabled People

771

and, in order to connect the Oculus Rift and the Move

to the machine, speciﬁc plug-ins and libraries were

used.

Unity. Unity

is a cross-platform game engine de-

veloped by Unity Technologies and used to develop

video games for PC, consoles, mobile devices and

websites. Recently, Unity Technologies made the

complete engine available for free including all fea-

tures, less source code and support.

Ofﬁcial Oculus Rift integration for Unity. In ad-

dition to the SDK, Oculus VR has released full sup-

port and integration for two well-known game en-

gines: Unity and Unreal Engine. By downloading

and installing the necessary plug-ins we were able to

directly use a connected Oculus Rift for any Unity

project without any issue.

PS Move plug-in for Unity: UniMove. UniMove

is an open-source Unity plug-in

, and set of C# bind-

ings that allows the usage of PS Move controllers

within a Unity game. The latest version works with

OS X 10.6+, Windows and Linux. It is based on top

of Thomas Perl’s PS Move API

. The plug-in is li-

censed under the GNU Lesser General Public License

and is still under development.

3.3 Overview of the Proposed Exergame

We propose a virtual reality game made to improve

the user’s awareness of road hazards and develop

his/her sense of direction in an urban environment

(e.g. a city). In this game the players must face a se-

ries of simple missions with increasing difﬁculty un-

der the supervision of an operator. Data relative to

each player’s performance is stored in a database, in

order to allow monitoring of the improvements of ev-

ery person. The proposed exergame can be seen as

a simulation of a real (and potentially dangerous) ac-

tivity in a secure and harmless context. A possible

application of proposed virtual reality game is to help

people with intellectual disabilities to learn about the

possible dangers of a city and to ﬁnd the way to a

known place in case they get lost. This is the reason

why a number of tests was run on this type of patients.

The Move is simply tied to the player’s thigh,

pointing downwards. The leg movement is then

tracked through the different accelerations along the

https://www.unity3d.com

http://www.copenhagengamecollective.org/projects/

unimove

http://thp.io/2010/psmove

Figure 1: Outline of the entire proposed system.

XY Z axes, and sent to the main application to be pro-

cessed into a realistic walking pattern (see Fig. 1).

The game has been created to be as much immersive

and noninvasive as possible. The possible actions in

the virtual world are simpliﬁed, so that the users can

focus only on the assigned tasks and the development

of their abilities (danger acknowledgment, sense of

direction). Our exergame lets people explore a Virtual

Reality environment simply by looking around (with

the Oculus Rift), and lets them move across this envi-

ronment by simulating a walk (with the PS Move).

The Virtual World. The world seen though the

Oculus Rift is just a simple urban environment, i.e.

a small city portion with buildings, streets, sidewalks,

lights and some minor details that can help in mak-

ing the overall experience more realistic. The graph-

ical interface is kept as simple as possible to avoid

interfering with the game’s immersivity: instructions,

messages and warnings are written in a textbox at the

bottom of the screen, and when a level is completed,

the message is supported by a single emoticon dis-

played at the center of the screen, which can be happy

or sad depending on the level’s result (Fig. 2).

Figure 2: Two screen examples: level cleared and level

failed. Note: the written text is in Italian language, since

the proposed exergame is for the use with Italian disabled

people.

The speakers mounted on the Oculus Rift can be

used to give additional instructions and warnings, but

they also generate sound effects typical of urban envi-

VISION4HCI 2016 - Special Session on Computer VISION for Natural Human Computer Interaction

772

Table 1: Organization of the ten levels of the proposed exergame.

Level Type Difﬁculty

1 Orientation Simple fork (go right or left)

2 Crossing Simple crosswalk without trafﬁc light nor vehicles

3 Orientation T intersection, single request

4 Crossing Crosswalk without trafﬁc light, with some vehicles

5 Orientation Two forks, two instructions

6 Crossing Crosswalk with trafﬁc light, mild trafﬁc

7 Orientation Two T intersections, two requests

8 Crossing Crosswalk without trafﬁc light, intense trafﬁc

9 Orientation Three intersections, three requests

10 Crossing Crosswalk with trafﬁc light, intense trafﬁc

ronments, such as moving vehicles, ambulance alarms

and general trafﬁc noise. This is done to emphasize

the sensation of immersivity for the player.

Gameplay. The proposed exergame offers two dif-

ferent challenges: general orientation and pedestrian

crossing. Each challenge is then divided into different

levels with increasing difﬁculty (see Table 1). In order

to proceed to the next level, the player has to complete

the previous ones, thus unlocking the new challenge.

The operator can keep track of the player’s progress

through the monitor, and have complete control over

which levels to start from, and which levels to repeat,

through a menu.

Orientation Levels. The orientation levels (Fig. 3)

take place in small city areas. Clear and simple in-

structions (go straight, turn left, turn right) are visual-

ized on the screen and/or sent through the Oculus Rift

speakers; if the player makes a mistake an error mes-

sage appears, the level is not cleared and it must be

repeated. For harder levels new elements are added

(intersections, U-turns), and a larger set of directions

is given to the player.

Crossing Levels. The crossing levels (Fig. 4) take

place along a road with various moving vehicles. The

presence of a crosswalk is always indicated on the

screen, while a trafﬁc light might or might not be

present, depending on the level. The vehicles are

managed by an artiﬁcial intelligence which takes into

consideration collisions and eventual trafﬁc lights.

Harder levels can feature unexpected anomalies in the

trafﬁc, such as ambulances and police cars.

In this type of challenge, the player is asked to

look around along the street in both directions before

making any move, otherwise the level results in auto-

matic failure. In a level with a trafﬁc light the player

must obviously wait for the green light ﬁrst, then look

Figure 3: In the orientation levels the player has to follow

the instructions (e.g. “Now turn right”) and reach the goal

following the right path. Note: the written text is in Italian

language, since the proposed exergame is for the use with

Italian disabled people.

Figure 4: In the crossing levels the player must look around

in both directions, and then cross the road when the light is

green.

around, and ﬁnally cross the street. In case a player

sees an ambulance or hears a siren, he/she must not

cross the road, even if the light is green.

Data Collection. The game has a login system to

keep track of whoever is playing, and it’s able to reg-

ister relevant data about the players’ performances.

An Oculus Rift based Exergame to Improve Awareness in Disabled People

773

These data are then stored and ordered into a database

that keeps a proﬁle for every player. The registered

data are: (i) starting/ending date/time of each test;

(ii) starting/ending time of each level; (iii) comple-

tion/failure of each level; (iv) reaction time after an

instruction has been given; (v) completion time for

each instruction, and whether the instruction has been

performed right or wrong. The entries are stored into

the database using a web service, therefore an Inter-

net connection is required to interact properly with the

system.

Game Options. Since everyone raises the legs in

slightly different ways, the operator can choose from

the main menu a series of options for each player, in

order to adjust the simulated walk parameters. The

sensitivity slider sets the upper threshold for the leg

lifting in order to maximize the number of recognized

steps. The stride length slider is the number of frames

in which the player moves forward when a single step

is recognized. The other sliders were added to adjust

the complexity of some levels, and they have no im-

pact on the user’s tracking.

3.4 The Walk Tracking Algorithm

In this paper, we present a fairly robust algorithm able

to detect when a player lifts a leg high enough to sim-

ulate the beginning of a step, and when he/she sets it

back down (presumably lifting the other leg) to end

the step. This way, a real person walking on the spot

translates to a player walking across the virtual city.

Fastening the PS Move to the Leg. The idea for the

tracking algorithm relies on the analysis of the leg’s

acceleration during a regular step, considering grav-

ity along one of the XY Z axes: since a standing per-

son normally starts walking with both legs perpendic-

ular to the ground (thus parallel to gravity, assuming

a ﬂat surface), and the Move’s sensor returns raw ac-

celerations along its own relative axes, the most logi-

cal choice seemed to be fastening the move along the

player’s thigh, pointing upwards or downwards, in or-

der to start the walk with one axis of the Move parallel

to gravity. The thigh was chosen over the calf because

the former performs a more regular movement during

the gait, which can pretty much be approximated to

a circular movement centered on the pelvis. For this

reason, the Move’s accelerations were used to com-

pute the angle at which the leg (or, more accurately,

the thigh) is lifted when a player performs a step.

Getting Acceleration Data from the Sensor. Con-

sidering the way the Move is tied to the thigh, when

the player stands still one of the axes of the controller

is parallel to gravity, and when the player lifts the leg

the angle between the thigh and gravity can be ap-

proximated using the variation of any of the three ac-

celerations registered by the sensor. For example, as-

suming that the Move is secured to the thigh as previ-

ously stated, and knowing how the three relative axes

of the controller are oriented with respect to the leg,

we could use accelerations along the Z and X axis

to compute our desired angle, since they are affected

from the rotational movement of the thigh. We chose

the Z axis and found out that the property Acceler-

ation.Z of the connected controller returns a normal-

ized value: 0 when the Move points to the ground (leg

down), and 1 when it points forward (leg up, thigh

parallel to the ground and perpendicular to gravity).

The resulting angles are then 0 degrees for a leg down

ad 90 degrees for a leg completely up.

Data Processing. One crucial issue of this game

is the robustness of the step recognition. The street

crossing levels put the player already on the edge

of the sidewalk, so any false step recognized by the

tracker immediately results into an actual step into

the virtual street. This can lead to unwanted failures

that alter the player’s record. For this reason an initial

check was added, overriding any other computation:

until the Z acceleration is sufﬁciently close to zero

(leg down), the step tracking is skipped completely.

This ensures that a player won’t start randomly walk-

ing in the virtual world until his initial position has

been conﬁrmed as idle. The step is recognized using

two parameters: the Z acceleration (zAcc), as men-

tioned above, and a boolean value named upLeg.

• zAcc is compared to two thresholds: the upper one

deﬁnes whether the leg has been lifted enough

to be the ﬁrst part of a step, while the lower

one (close to zero) deﬁnes whether the leg is low

enough for the step to be considered completed.

• upLeg’s purpose is to deal with the obvious con-

ﬂicts generated by the acceleration alone. When

we estimate the position of a leg, we must take

into account its previous position as well. A suc-

cessful ”leg up” implies that zAcc has surpassed

the upper threshold from below, hence a check on

upLeg must be done, and it must be false. Sim-

ilarly, for a ”leg down” zAcc has to be under the

lower threshold, and upLeg must be true.

Without any check on the previous leg state, even

standing still would generate an inﬁnite walk, because

zAcc would consistently be under the lower threshold,

thus breaking the algorithm. In a real case scenario

the leg-up condition is much more critical than the

VISION4HCI 2016 - Special Session on Computer VISION for Natural Human Computer Interaction

774

leg-down condition. While a leg down equals to 0

degrees most of the time, every person, when walk-

ing, lifts his legs by a certain amount, which is not the

same for everyone (Fig. 5).

Figure 5: Walking pattern of an ordinary person.

It’s easy to see that no one lifts his/her thigh up to

90 degrees when walking, and some simple tests re-

vealed that even 45 degrees is often too much to set

as the upper threshold for the step recognition. How-

ever, ﬁnding a seemingly right angle was not enough,

because some players could always lift their leg even

less than the testers, and if the leg-up is not recog-

nized, the entire step gets skipped. The solution to

this problem was setting the upper threshold dynami-

cally from the game’s main menu. This way, the sys-

tem can be calibrated runtime for every player, and

the settings can be saved into the database for future

use. With these precautions we were able to ensure a

smooth and robust tracking, without losing steps nor

generating false ones.

Player’s Movement in the Virtual World. In the

virtual city, the distance covered by a single step

depends on two factors: the number of frames in

which the player moves forward, and the amount

of space covered in each frame. The actual func-

tion for the player’s movement within a frame is:

Move( drifter.facingDirection.forward* drifter.speed*

Time.deltaTime)

It is simple to understand: the drifter is the player,

and his facing direction is computed directly from the

Oculus Rift (the player always walks towards his fac-

ing direction), then there is Time.deltaTime, which is

the timespan of a single frame, and ﬁnally there is

the walking speed. This parameter has to be reﬁned

test after test, based on the player’s height and the

size of the roads, in order for the walk to seem as

realistic as possible without the sensation of running

(crossing the entire road in a single step). Since the

speed over a single frame is de facto instantaneous,

the Move() function causes just a minimal shift. A

second parameter is then needed to create a movement

that takes place over the span of many frames. This

variable, named walkFrameCount, deﬁnes the num-

ber of frames before a player stops moving after a

successful step.

The frame count variable has actually a double

use: to be realistic, the simulated walk must not be

jerkily at all, thus between the leg-up and the leg-

down the player must absolutely continue his move-

ment. The timespan of a virtual step depends heavily

on how fast a person walks on the spot, so another

custom option for each player in the main menu was

added. As a counter, walkFrameCount must be ini-

tialized to the desired value at every successful step.

While the player is moving, every frame decreases the

counter until it reaches zero and then stops the shift.

If another step is done before the counter reaches the

end, it gets reinitialized and the shift doesn’t stop. If

the player performs several steps in a row, and the

frame count is high enough to cover both phases of

the step, the result will be a continuous movement to-

wards the facing direction. With a good balance be-

tween speed and walkFrameCount, the player can ex-

perience a realistic shift when walking in the virtual

city, and doesn’t need to worry about his avatar freez-

ing mid-step.

4 RESULTS

4.1 Subjects

The proposed exergame has been test with 6 subjects

between 14 and 18 years old. They were recruited

on a voluntary basis, and their legal representatives

agreed on a Consent Form, in which the description of

the experiments, together with explicit statements re-

garding ethical considerations in the recruitment and

treatment of all subjects, was included. The research

will conform to the ethical standards laid down in the

1964 Declaration of Helsinki that protects research

subjects.

The proposed game was designed mainly to aid

people with cognitive deﬁcits, so a number of tests

on a selected sample of patients has been run. More

in detail, the patients were divided in two groups:

one with people affected by Down Syndrome, and one

with people affected by Fragile X Syndrome. Clearing

all the levels with high scores wasn’t the only goal

of the project, a good video game must also give a

great experience to stimulate the users to keep play-

ing, especially with such a heterogeneous hardware

interface. For this reason, there were some possible

issues to deal with:

Comfort. We propose a wearable system, so in or-

der to give the players the best experience it must be

An Oculus Rift based Exergame to Improve Awareness in Disabled People

775

Table 2: Data gathered during the experimental session for a sample of patients.

Player Clearing time Mission type Level Day/night Reaction time Error Error type

(mission) (task)

B. 30.169 s Orientation 2 Day 1.973 s No -

C. 18.11 s Orientation 4 Night 4.991 s Yes 1

G. 31.813 s Crossing 3 Night 8.802 s Yes 4

L. 21.169 s Orientation 8 Night 3.311 s No -

M. 15.513 s Crossing 5 Day 9.259 s Yes 2

N. 53.67 s Orientation 2 Day 21.977 s No -

comfortable to wear, even for prolonged periods of

time. The Oculus Rift is a fairly heavy headset, and

a continuous usage can be tiresome for both the neck

and the eyes. The PS Move on the other hand must be

tied ﬁrmly to the leg, to avoid discrepancies in the ac-

celerations’ processing, but not too tightly, as it would

risk hurting the patient.

Immersivity. Even with all the expedients done to

ensure maximum immersivity, patients with cognitive

deﬁcits might not feel comfortable with the virtual

reality environment and the interaction with the ex-

ergaming devices. If that were to happen they proba-

bly wouldn’t even be stimulated to walk and explore

the virtual world, thus defeating the purpose of the en-

tire game being based on Human-Computer-Interface

and Virtual Reality.

Focus. Instructions for clearing the levels can be re-

ceived on-screen or through the Oculus Rift speak-

ers. As previously mentioned, the graphical interface

was kept as simple as possible to avoid obstruction of

the virtual experience, but it must be at the same time

clear end effective, because if a patient doesn’t even

notice a new instruction is available, that will greatly

affect his/her reaction and completion times in the test

results.

4.2 Experimental Session

Experiments were run with different combinations of

game parameters. Some videos of the experiments

can be found at the following links

6 7 8

. For example,

in addition to the level type and difﬁculty, the operator

could choose the time of the day in which the virtual

city was displayed. This was added to give the play-

ers awareness about exploring a city in different light

conditions. In spite of the possibility of the aforemen-

tioned technical problems, the tests went surprisingly

https://youtu.be/jy67h2hug1I

https://youtu.be/ujUF9JeJ1EU

https://youtu.be/vQEZIVro8 8

well. The patients collaborated without any issue, and

the general consensus was that they truly enjoyed the

game mechanics and the possibility to explore an in-

teractive world. From the gaming point of view this

project was surely a success. The human-computer

interface worked ﬂawlessly most of the time, register-

ing steps in the right way without disrupting the gam-

ing experience. An example of the data gathered dur-

ing the experimental session the simulation is shown

in Table 2.

The tests yielded good results as well as bad ones:

some patients had more difﬁculties clearing the harder

levels, but everyone was at least able to clear the

basic ones within a few tries. The time of the day

didn’t seem to inﬂuence the performance, and while

the level difﬁculty surely did, the most skilled players

didn’t have any particular issue with the increasing

number of instructions, nor with the trafﬁc intensity.

The reaction times per instruction changed a lot, even

within the same player. Depending on how complex

the indication was, some players performed their ac-

tions in up to 30 seconds, while the lowest reaction

times ranged between 2 and 4 seconds. An error type

ﬁeld (i.e. 1: risk of being hit by a car; 3: cross walk

when trafﬁc light is yellow; 3: cross walk when trafﬁc

light is red; 4: cross walk outside zebra crossing) was

added to keep track of the causes of every error, for a

more accurate analysis of the players’ capabilities.

5 DISCUSSION AND

CONCLUSION

The tests run during the debugging phase and the ex-

periments run on the users conﬁrmed the potentialities

of the proposed exergame:

• The virtual world was perfectly perceived by the

players despite being simpliﬁed with just the es-

sential details;

• The navigation was realistic and the tracking al-

gorithm was robust enough: the player could feel

VISION4HCI 2016 - Special Session on Computer VISION for Natural Human Computer Interaction

776

himself walking in the city as if the roads were

real;

• The system was responsive: there were no delays

when looking around with the Oculus Rift, nor

when walking with the Move, everything was felt

real-time;

• The patients had fun and this is a crucial aspect for

a video game, even without taking into account

the test results: fun must be something that differ-

entiates a game from a simple interactive applica-

tion for medical purposes;

The concept of this game could be used for new and

different applications in both video game and medical

ﬁelds, with the use of new interfacing devices other

than the Oculus Rift and the Move.

REFERENCES

Chan, S., Conti, F., Salisbury, K., and Blevins, N. H. (2013).

Virtual reality simulation in neurosurgery: technolo-

gies and evolution. Neurosurgery, 72:154–164.

Deutsch, J., Borbely, M., Filler, J., Huhn, K., and Guarrera-

Bowlby, P. (2008). Use of a lowcost, commercially

available gaming console (Wii) for rehabilitation of

an adolescent with cerebral palsy. Physical Therapy,

88(10):1196–1207.

Deutsch, J., Robbins, D., Morrison, J., and Bowlby, P.

(2009). Wii-based compared to standard of care bal-

ance and mobility rehabilitation for two individuals

post-stroke. In Proceedings of the Virtual Rehabili-

tation International Conference, pages 117–120.

Hoffman, H., Meyer, W., Ramirez, M., Roberts, L., Seibel,

E., Atzori, B., Sharar, S., and Patterson, D. (2014).

Feasibility of articulated arm mounted Oculus Rift vir-

tual reality goggles for adjunctive pain control dur-

ing occupational therapy in pediatric burn patients.

Cyberpsychology, Behavior, and Social Networking,

17(6):397–401.

Hsu, J., Thibodeau, R., Wong, S., Zukiwsky, D., Cecile,

S., and Walton, D. (2011). A ”Wii” bit of fun: The

effects of adding Nintendo Wii Bowling to a standard

exercise regimen for residents of long-term care with

upper extremity dysfunction. Physiotherapy Theory

and Practice, 27(3):185–193.

Hurkmans, H., Berg-Emons, R., and Stam, H. (2010). En-

ergy expenditure in adults with cerebral palsy playing

Wii Sports. Archives of Physical Medicine and Reha-

bilitation, 91(10):1577–1581.

Joo, L., Yin, T., Xu, D., Thia, E., Chia, P., Kuah, C., and

He, K. (2010). Feasibility study using interactive com-

mercial off-the-shelf computer gaming in upper limb

rehabilitation in patients after stroke. Journal of Re-

habilitation Medicine, 42(5):437–441.

Kliem, A. and Wiemeyer, J. (2010). Comparison of a tra-

ditional and a video game based balance training pro-

gram. International Journal of Computer Science in

Sport, 9:80–91.

Kwon, J. H., Powell, J., and Chalmers, A. (2013). How

level of realism inﬂuences anxiety in virtual reality en-

vironments for a job interview. International Journal

of Human-Computer Studies, 71(10):978–987.

Parker, J., Baradoy, G., and Katz, L. (2011). Using virtual

reality technology and biometric interfaces in obesity

reduction. Canadian Journal of Diabetes, 35(2):187.

Sharples, S., Cobb, S., Moody, A., and Wilson, J. R. (2008).

Virtual reality induced symptoms and effects (vrise):

Comparison of head mounted display (hmd), desktop

and projection display systems. Displays, 29(2):58 –

69.

Sohnsmeyer, J., Gilbrich, H., and Weisser, B. (2010). Effect

of a 6-week intervention with an activity-promoting

video game on isometric muscle strength in elderly

subjects. International Journal of Computer Science

in Sport, 9(2):75–79.

Solari, F., Chessa, M., Garibotti, M., and Sabatini, S. P.

(2013). Natural perception in dynamic stereo-

scopic augmented reality environments. Displays,

34(2):142–152.

Tanaka, K., Parker, J., Baradoy, G., Sheehan, D., Holash,

J., and Katz, L. (2012). A comparison of exergam-

ing interfaces for use in rehabilitation programs and

research. The Journal of the Canadian Game Studies

Association, 6(9):69–81.

Ukai, K. and Howarth, P. (2008). Visual fatigue caused

by viewing stereoscopic motion images: Background,

theories, and observations. Displays, 29(2):106–116.

Waterworth, J. A., Waterworth, E. L., Mantovani, F., and

Riva, G. (2010). On feeling (the) present: An evolu-

tionary account of the sense of presence in physical

and electronically mediated environments. Journal of

Consciousness Studies, 17(1-2):167–188.

Wollersheim, D., Merkes, M., Shields, N., Liamputtong, P.,

Wallis, L., Reynolds, F., and Koh, L. (2010). Phys-

ical and psychosocial effects of Wii video game use

among older women. International Journal of Emerg-

ing Technologies and Society, 8(2):85–98.

An Oculus Rift based Exergame to Improve Awareness in Disabled People

777