INVESTIGATING THE INFLUENCE OF COMMUNICATION AND

INPUT DEVICES ON COLLABORATION IN VIRTUAL

ENVIRONMENTS

Anastasiia Beznosyk, Chris Raymaekers, Karin Coninx, Peter Quax and Wim Lamotte

Hasselt University — tUL — IBBT, Expertise Centre for Digital Media

Wetenschapspark 2,B-3590 Diepenbeek, Belgium

Keywords:

Collaborative virtual environments, 3D manipulation, Input devices.

Abstract:

As collaborative virtual environments (CVEs) are becoming more popular for both entertainment and profes-

sional activities, it is important to know which factors inﬂuence the collaboration between participants. This

paper investigates two aspects of interaction in the context of collaboration between two users. A puzzle-

solving task in a basic interactive virtual environment was used as a case study. The inﬂuence of voice com-

munication was assessed in a user experiment. Furthermore, as different I/O devices can be used with this

kind of applications, the fact whether both participants are using the same device or not, was also checked.

From this study, we can conclude that the inclusion of voice communication is important when working on

the same task in a CVE, since this allows the participants to explicitly divide the work. On the other hand,

the usage of the same kind of device by both users does not signiﬁcantly inﬂuence the collaboration. The

availability of different devices shows not to be a problem for the acceptance of CVEs, and therefore it is not

necessary to impose a certain device, which is obviously preferable, especially when dealing with home users.

1 INTRODUCTION

The necessity for communication and information

sharing in virtual environments (VEs) caused the

appearance of collaborative virtual environments

(CVEs). Massive Multiplayer On-line Role Play-

ing Games and virtual communities, such as Sec-

ond Life (Linden Labs, http://www.secondlife.com,

2003), have contributed signiﬁcantly to this success.

The variety of computational resources nowadays

provides a wide range of possibilities for the devel-

opers of virtual worlds. The potential of these re-

sources leads to the question of their effective combi-

nation into heterogeneous set-ups, in such a way that

the tasks between participants can be divided accord-

ing to their abilities.While single-user interaction has

been studied at great lengths (Bowman et al., 2005),

the problem of enabling detailed interaction using a

heterogeneous set of input devices is non-trivial to

solve. To achieve productivity and efﬁciency of col-

laboration a detailed investigation of different device

combinations is needed.

Another aspect that has received little attention is

the effect of communication while working in a 3D

environment (Otto and Roberts, 2003; Ullah et al.,

2009). Several studies have shown its importance to

increase the degree of presence within the environ-

ment. These studies, however, do not take the pro-

ductivity of the collaboration into account.

We have investigated the inﬂuence of heteroge-

neous interaction when collaborating while perform-

ing highly interactive tasks. More speciﬁcally, the ex-

periment studies the inﬂuence of unequal conditions

between participants, as they are collaborating using

different input devices. Although the performance

and feeling of collaboration can be considered as the

task dependent characteristics we believe that the out-

come obtained in this particular case study can be ap-

plied to a wider range of tightly coupled interactive

tasks.

2 RELATED WORK

Most current CVEs support a high level of presence

and immersion, however, the interaction is reduced to

voice communication and common navigation. De-

341

Beznosyk A., Raymaekers C., Coninx K., Quax P. and Lamotte W. (2010).

INVESTIGATING THE INFLUENCE OF COMMUNICATION AND INPUT DEVICES ON COLLABORATION IN VIRTUAL ENVIRONMENTS.

In Proceedings of the International Conference on Computer Graphics Theory and Applications, pages 341-346

DOI: 10.5220/0002822403410346

 SciTePress

tailed task performance has received less focus.

Some previous studies examined such topics as

presence and co-presence (Schroeder et al., 2001). In

other examples (Alhalabi and Horiguchi, 2001; Duval

et al., 2006) collaboration was supported between 2

users with equal performance abilities. Another ex-

ample of interaction, that is homogeneous with re-

spect to the I/O devices, was presented by Pinho et

al. (Pinho et al., 2008) but in this study responsibil-

ities were divided between the participants so they

were able to complete the task only by cooperating.

Performance, leadership and presence have been

identiﬁed as key issues in virtual environments (Hel-

dal et al., 2005). Some of these factors depend on the

type of the virtual environment and the tasks to be per-

formed, others on computational resources and avail-

able devices. For example, experiments (Steed et al.,

1999) have proved that the person who had the highest

degree of feeling to be present in the VE was singled

out as the leader. In the meantime there was little re-

search that investigated the inﬂuence of communica-

tion between participants on collaboration (Otto and

Roberts, 2003). There is also a lack of studies inves-

tigating the absence of communication and possible

negative reactions due to the ability to talk.

The contribution of our work is the combination

and comparison of different setups of input devices.

Although several studies are known where different

devices were combined, they included similar kinds

of devices (McLaughlin et al., 2003). We com-

pare not only performance between different device

types (relative and absolute) within a homogeneous

setup, but we also show the heterogeneous perfor-

mance when everyone is using a different device.

3 EXPERIMENT

In order to assess the inﬂuence of communication

and the I/O devices used, a user experiment was con-

ducted. Pairs of participants had to collaborate in a

highly interactive puzzle-solving task. To investigate

the effect of a homogeneous setup versus a hetero-

geneous setup, a SpaceMouse and a Phantom device

were chosen. While the Phantom device has an abso-

lute mapping of its position and orientation to the 3D

cursor, the SpaceMouse is a relative device.

3.1 Participants

Six female and twenty-six male unpaid volunteers,

ranging in age between 23 and 39 participated in the

experiment. They were randomly divided into 16

pairs. All test persons had a computer science back-

ground. All but one participant were right-handed and

used their dominant hand to operate the devices. Most

participants had no experience working with the in-

put devices involved. The others only had experience

from participating in other user tests, where the same

devices were involved.

3.2 Apparatus

For the experiment, two desktop computers were con-

nected over a LAN. For output two 19” monitors were

used. As indicated earlier, a Phantom device and a

SpaceMouse were used for input. Participants were

sitting in the same room quite close to each other but

they were not able to see the partner’s screen, as can

be seen in ﬁgure 1.

Figure 1: Experimental setup.

3.3 Design

For selection and manipulation within the 3D envi-

ronment, the virtual hand technique (Bowman et al.,

2005) was chosen. The input devices were repre-

sented by cones of different colors.

A between-subjects design was used for the pres-

ence or absence of voice communication. The de-

vice order was counterbalanced using a Latin square

design. This resulted in four puzzles that had to be

solved by each pair: two using homogeneous interac-

tion and two with heterogeneous interaction. The de-

pendent variables were the completion time for each

group and the amount of correctly placed blocks for

each participant.

3.4 Procedure

The puzzle-task consisted of collecting 12 pieces, rep-

resenting a picture, where one was already provided

as visual guide (see ﬁgure 2). Each session consisted

of a new puzzle. The puzzles were assigned to all

pairs in the same order and had the same level of com-

plexity. The participants saw the completed picture in

front of them during the experiment. Puzzle pieces

were represented as cubes, with the picture on one

GRAPP 2010 - International Conference on Computer Graphics Theory and Applications

342

face, dispersed in the 3D environment. Start and Fin-

ish buttons had to be clicked when they were ready

to perform the task and as soon as the task had been

completed. The selected puzzle piece was highlighted

for both participants to avoid same cube selection.

In order to investigate the inﬂuence of commu-

nication, half of the pairs were instructed to talk to

each other. As not all participants spoke the same lan-

guage as ﬁrst language, all communication was held

in English. Although the participants were not native

speakers, we believe this did not signiﬁcantly inﬂu-

ence the results as all of them were used to speak En-

glish on a regular basis.

Figure 2: Experimental task in the VE.

In the beginning of the experiment, participants

were allowed to practice with both devices. Once the

training was completed, the four puzzle-solving ses-

sions were performed. After each session, the partic-

ipants had to ﬁll in a questionnaire, based on (Steed

et al., 1999).

4 ANALYSIS AND RESULTS

4.1 Communication

In order to assess the users’ performance, we ana-

lyzed how the ability to talk inﬂuenced the total time

to perform the task. Against our expectations the task

was completed quicker in all 4 sessions for the pairs

who were not allowed to talk. In table 1 the perfor-

mance results of homogeneous (HM) and heteroge-

neous (HT) setups are compared (M – mean, SD –

standard deviation).

Table 1: Total time (sec).

Communication

No communication

M SD M SD

Homogeneous interaction

206.01

89.43

182.30

64.97

Heterogeneous interaction

235.64 133.75 166.75 58.50

Total time on task was analyzed using a linear

mixed model. We found a signiﬁcant main effect

of communication on total time (F

1,58

= 4.768, p =

0.03). This indicates that people spent signiﬁcantly

more time on a task when they were able to talk to the

partner. This occurred regardless of the devices used,

as no interaction effect was found between the setup

and communication.

One of the characteristics that test persons had to

evaluate was the convenience with a certain device.

Figure 3 shows that using the same device setup under

different conditions changed the feeling of the partic-

ipants. For example, when they were allowed to talk,

the mean value of “convenience” was higher. This

can be explained by the fact that when people were al-

lowed to communicate they were able to ask the other

person for help so they did not feel themselves being

aside from the performance.

Figure 3: Convenience with the device, for heterogeneous

(HT) and homogeneous (HM) setups.

The generalized linear model (GLM) analysis has

shown that for the no-communication case there is a

signiﬁcant main effect of the convenience with device

on the total amount of blocks placed by a participant

(χ

(1) = 5.671, p = 0.017). For the case where com-

munication was allowed, there is a main effect of the

device convenience on the total time (F

1,27

= 11.094,

p = 0.003).

The results of the experiment show that the abil-

ity to communicate in a CVE increases the feeling of

collaboration independent of the setup (see table 2).

Table 2: Feeling of collaboration.

Communication

No communication

M SD M SD

Feeling of collaboration

7.19

1.36

5.59

2.27

We therefore come to the following conclusion.

When it is important to achieve a better feeling of

“working together”, communication plays an essen-

tial role. But if it is crucial to perform the task quicker

and divide it between the participants, the ability to

communicate can decrease the productivity and efﬁ-

INVESTIGATING THE INFLUENCE OF COMMUNICATION AND INPUT DEVICES ON COLLABORATION IN

VIRTUAL ENVIRONMENTS

343

ciency of the collaboration.

For the sessions with communication, the linear

mixed model analysis shows a main effect of the part-

ner’s share to the task solving on the “feeling of col-

laboration” (F

7,11

= 4.327, p = 0.015). We found

that the contribution to the placing of the cubes is

not signiﬁcant for this case. The partner’s conve-

nience with the device has main effect on collabo-

ration (F

9,20

= 2.445, p = 0.046), which indicates

that people felt to collaborate signiﬁcantly more when

their partners were comfortable with the devices.

For the case where no conversation was allowed,

the own contribution to the cube placing is signif-

icant for the level of collaboration (F

1,27

= 13.099,

p = 0.001).

Another relevant result that was obtained by com-

parison of the performance under two conditions

shows that in all sessions the mean values of contribu-

tion to the task by the participants were higher when

they had the chance to talk to each other and discuss

the task (see ﬁgure 4).

Figure 4: Contribution to the task, when using a Space-

Mouse (M) or a Phantom device (P).

Amongst others we wanted to determine the fac-

tors that deﬁne leadership when collaborating. One

of our initial expectations was that the person would

be considered as a leader more when he/she performs

more active and places more blocks to the correct po-

sition in both cases. But for the communication cases

it turned out that people who were assessed as being

talkative were seen as the leaders, rather than partic-

ipants who performed more. Our analysis led to the

conclusion that when people are not able to commu-

nicate with each other, they take the lead when they

feel comfortable to operate within the environment

and this can be achieved by using suitable devices.

Participants were also asked to estimate to what

extent they considered themselves as a leader. Fur-

thermore, we asked to evaluate the extent of the lead-

ership of the partner. In our opinion it is interesting to

compare how these evaluations differ from each other

under different conditions.

In the cases where communication was allowed,

the persons who operated a SpaceMouse had the same

idea about their leadership as their partners. In con-

trast, when the person was using a Phantom device,

he/she tended to overrate own leadership value com-

paring to the one given by their partner. For the cases

without communication, there is little difference be-

tween the self evaluation and the partner’s rating.

Comparing the correlation between leadership and

convenience with the device under the two conditions

the following can be observed: for the communication

case the leadership is deﬁned mostly by the commu-

nication, while for the no-communication cases, the

comfort of device usage is more inﬂuential

The performed analysis has shown that the corre-

lation between the device convenience and leadership

is stronger for no-communication cases. For the com-

munication cases, there is an input of the communica-

tion to the correlation.

Mixed model analysis has shown no main ef-

fect of the contribution characteristics on the lead-

ership value for the communication case. Contrary to

this, for no-communication cases, there is a signiﬁ-

cant main effect of blocks placed by the participant

1,24

= 5.064, p = 0.34) and his contribution to the

task solving (F

1,24

= 6.247, p = 0.020) on the level of

his perceived leadership.

4.2 Homogeneous and Heterogeneous

Setups

Our research was organized with the aim to inves-

tigate the potential of unequal interaction (heteroge-

neous setup) within one collaborative environment.

One of the main goals we wanted to achieve was to

see the possibilities and drawbacks of creating het-

erogeneous collaborative environments where every-

one can perform speciﬁc tasks to achieve a common

aim. Although we did not perform such a separation

in this experiment, results have shown that in com-

munication cases, participants divided their roles by

themselves.

Analyzing the heterogeneous and homogeneous

setups, based on a linear mixed model, we found no

signiﬁcant difference for the performance time, but

there is a main effect of convenience with the device,

1,57

= 4.050, p = 0.049 for participant #1 and F

1,57

= 23.224, p<0.001 for participant #2. There was no

signiﬁcant interaction effect between 2 devices. This

result provides us with the ability to design collabo-

rative environments where the use of certain devices

does not limit other device types that could be in-

volved.

Using a GLM analysis of the effect of different

conditions on the total amount of blocks placed cor-

rectly, we found a main effect of one of the device

GRAPP 2010 - International Conference on Computer Graphics Theory and Applications

344

types (χ

(1) = 3.869, p = 0.049).This indicates that a

pair placed signiﬁcantly more blocks if at least one of

the devices was comfortable to use.

The change from a homogeneous device combi-

nation to a heterogeneous one during communication

session does not bring any signiﬁcant result (Fig. 5).

However, when no communication is present, there

is a tendency to improve the feeling of collaboration

and increase the efﬁciency in the heterogeneous setup

(Fig. 6). Although the available activities of both par-

ticipants were the same, they felt more comfortable

when they could rely on their partner using a different

device.

Figure 5: Effect of the device on the collaboration in the

communication case.

Figure 6: Effect of the device on the collaboration in the

no-communication case.

According to the total time to perform the task,

we can see that the heterogeneous setup reduces

the completion time compared to the SpaceMouse—

SpaceMouse case. It takes, however, longer than the

Phantom—Phantom case.

Figure 7 shows the dependence between the share

in task solving and the type of the device used.

Because the SpaceMouse was considered by most

participants as a more complicated device to use, the

SpaceMouse—SpaceMouse case is rather changing

the character of the trend (jumping from min extreme

to max several times). The homogeneous case with

the Phantoms shows that the contribution is quite

high and remains almost the same during the whole

Figure 7: Share in task solving.

experiment. In the heterogeneous case, the partici-

pant using the Phantom has a higher share in the task

solving compared to the same device in the homoge-

neous case. The heterogeneous case SpaceMouse—

Phantom shows that the participant using the Space-

Mouse felt as having contributed less in solving the

task. Analysis has shown that users with less com-

fortable devices did not enjoy the collaboration. They

were not always able to follow their partners. This

is the main drawback of unequal collaboration. Such

consequences should orient the developers of collab-

orative environments towards more balanced cooper-

ation. The development of appropriate tasks, where

the diversity of abilities can be beneﬁcial for all the

participants and involve everyone at the same level of

cooperation, is one possible solutions.

For this reason, it is also important to investigate

the correlation between the device usage and the con-

tribution to the block placing. Figure 8 shows this

correlation for the cases, where communication was

present.

Figure 8: Block placing in the communication case.

As can be seen from the graph for the homoge-

neous case, most of the values remain within a cer-

tain range and have little data spread. For the het-

erogeneous sessions more extreme values can be ob-

served, especially in the case when the participant

was using the SpaceMouse. But it is also necessary

to keep in mind that during the communication case

the “placing cubes” variable did not always show the

reality, as some pairs divided the work: one part-

ner would orientate all puzzle pieces, while the other

would place them in the puzzle. That is why we also

INVESTIGATING THE INFLUENCE OF COMMUNICATION AND INPUT DEVICES ON COLLABORATION IN

VIRTUAL ENVIRONMENTS

345

present the same analysis for “placing cubes” in the

no-communication case (see ﬁgure 9).

Figure 9: Block placing within no-communication case.

The graph of no-communication case shows a

more balanced trend with less extreme values. There

is also not much data spread for the homogeneous

case which was caused by more balanced collabora-

tion.

The results obtained in this comparative study of

homogeneous and heterogeneous setups can be used

in 2 ways. The ﬁrst conclusion that can be drawn here

is that unequal interaction (in terms of I/O devices)

can decrease the efﬁciency and contribution of one of

the participants, and cause an uncomfortable experi-

ence for these participants in a CVE. However as, in

reality, the device that is most suited for performing

a task is not always available, we can conclude that

a heterogeneous setup still allows users to work in an

efﬁcient manner. On the other hand, we can think

about other types of tasks to be performed in a CVE,

where participants have limited abilities and should

perform different tasks to achieve the common goal.

5 CONCLUSIONS

The results obtained in this case study have an im-

portant impact for the implementation of collabora-

tive virtual environments. One implication is that in-

clusion of communication is important when working

together in a virtual environment, since this allows

the participants to explicitly divide the work. Com-

munication also increases the feeling of collaborating

together but at the same moment decreases the per-

formance. Another important result is that the com-

bination of different devices does not signiﬁcantly in-

ﬂuence the collaboration as no interaction effect of

the devices that can be combined was found. To in-

crease the efﬁciency of heterogeneous collaboration

we can think of other activities to be performed in

the VE where every participant can supplement each

other performance. Of course further research is nec-

essary to explore other aspects of successful collabo-

ration.

ACKNOWLEDGEMENTS

The research described in this paper is directly funded

by Hasselt University through the BOF framework.

The authors would like to thank Lode Vanacken

for his assistance with the implementation and Roel

Braeken of the UHasselt Center for Statistics for help-

ing to choose the statistical tests.

REFERENCES

Alhalabi, M. O. and Horiguchi, S. (2001). Tele-handshake:

Cooperative shared haptic virtual environment. Re-

search report, pages 1–12.

Bowman, D. A., Kruijff, E., LaViola, J. J., and Poupyrev,

I. (2005). 3D User Interfaces, Theory and Practice.

Addison-Wesley.

Duval, T., Lecuyer, A., and Thomas, S. (2006). Skewer: a

3d interaction technique for 2-user collaborative ma-

nipulation of objects in virtual environments. In 3DUI

’06: Proceedings of the 3D User Interfaces, pages 69–

72, Washington, DC, USA. IEEE Computer Society.

Heldal, I., Steed, A., and Schroeder, R. (2005). Evaluating

collaboration in distributed virtual environments for a

puzzle-solving task. In Proceedings of the 11th Inter-

national Conference on HCI.

Linden Labs, http://www.secondlife.com (2003).

McLaughlin, M., Sukhatme, G., Peng, W., Zhu, W., and

Parks, J. (2003). Performance and co-presence in het-

erogeneous haptic collaboration. Haptic Interfaces for

Virtual Environment and Teleoperator Systems, Inter-

national Symposium on, pages 285 – 291.

Otto, O. and Roberts, D. (2003). Importance of com-

munication inﬂuences on a highly collaborative task.

Distributed Simulation and Real-Time Applications,

IEEE International Symposium on, page 195.

Pinho, M. S., Bowman, D. A., and Freitas, C. M. (2008).

Cooperative object manipulation in collaborative vir-

tual environments. In J. Braz. Comp. Soc., volume 14,

pages 54–67.

Schroeder, R., Steed, A., Axelsson, A.-S., Heldal, I.,

Asa

Abelin, Widestr

om, J., Nilsson, A., and Slater, M.

(2001). Collaborating in networked immersive spaces:

as good as being there together? Computers & Graph-

ics, 25(5):781 – 788.

Steed, A., Slater, M., Sadagic, A., Bullock, A., and Tromp,

J. (1999). Leadership and collaboration in shared

virtual environments. In VR ’99: Proceedings of

the IEEE Virtual Reality, page 112, Washington, DC,

USA. IEEE Computer Society.

Ullah, S., Otmane, S., Richard, P., and Mallem, M. (2009).

Human performance in cooperative virtual environ-

ments : the effect of visual aids and oral communi-

cation. IJVR, 8(4):79–86.

GRAPP 2010 - International Conference on Computer Graphics Theory and Applications

346