A Study of Virtual Environments for Enterprise Collaboration Aditya Zutshi ∗ Geetika Sharma † TATA Consultancy Services Innovation Labs, Delhi
Abstract
important for collaboration and compare three specific virtual environments. We highlight areas of improvement in the current state of the art VEs and suggest some good practices to follow in setting up and using VEs. This paper is organised as follows. In section 2, we give a brief background of VEs. We describe our study and its results in section 3 and conclude in section 4.
Globally distributed workforces in enterprises are commonplace these days due to the increasing popularity of the off-shore business model. One of the challenges faced by these enterprises is enabling effective collaboration between geographically separated teams while keeping operational costs to a minimum. Although a number of tools have been developed for collaboration, online virtual environments (VEs) open up huge possibilities. In this paper, we present the results of a pilot study we conducted on the usability and acceptability of some popular virtual environments for collaboration within an enterprise. Based on our evaluation, we analyse where the current class of VEs lack and prescribe some good practices for their use.
2 Online Virtual Environments A virtual environment (VE) is a shared, persistent computer simulated space into which multiple users can interact with each other in real-time [Bartle 2004]. The environment is governed by a set of rules or physics, that allow users to cause changes in it. As the user navigates through the environment a view of the current state of the world is presented to him. The user’s view started out, historically, from being textual (e.g. MUDs [The MUD Connector ]), 2.5D (e.g. EverQuest [EverQuest ]) to full 3D e.g.Ultima Online [Ultima Online ], Second Life [Second Life ]). Textual environments provide a text-based description of the world and the people in it. The text may be marked up with colours to add meaning. A 2.5D world allows 2D user navigation with a fixed isometric view that creates an impression of 3D while a full 3D world allows the user 3D navigation and to view the world from almost any orientation. Since our study is concerned with 3D VEs, we focus on these in the remainder of the paper.
Keywords: user study, virtual environments, enterprise collaboration
1
Introduction
With offices worldwide, and customers and business partners spread across the globe, organizations are trying to find a way of collaborating that is as productive as face-to-face meetings and capable of supporting groups of large sizes. A large number of collaboration tools ranging from web conferencing to integrated suites that can create knowledge repositories and enable document sharing, meeting and conversations [Weiseth et al. 2006] are available today.
The current class of online 3D VEs allows multiple users to connect over a network in a shared space that is built using 3D graphical objects. The objects may be scripted to behave in a desired manner. Users in the environment are represented as avatars which are customisable 3D graphical models, usually with a human-like appearance. The user’s interaction in the virtual space is through the avatar. To move around in the space, the user navigates his avatar. The user’s view of the space is created by setting up a camera with respect to the avatar. This could be, for example, a first person view in which the camera position and orientation coincides with that of the avatar’s head or a third person view in which the camera is above and behind the avatar’s head. Avatar’s may be animated to perform gestures and actions to make the interaction more meaningful and engaging.
Online virtual environments (VEs) are becoming increasingly popular after the unprecedented success of Second Life [Second Life ]. Although a large share of users log into VEs for gaming or socialising, enterprises are looking for ways to leverage this technology for increasing productivity and reducing costs. VEs have a potential for improving collaboration, [Brown and Bell 2006], as they provide a shared space into which multiple users can log in from different parts of the world and interact with each other in real-time. While this interaction can be for fun, it can also be for work. The commonly used collaboration tools such as text and VoIP chat, video sharing, document sharing and so on, can be integrated in a 3D space in a manner that provides for better organisation and comprehension of different data streams. Also, since the real-world is 3D, our minds are wired to naturally interact and process information in 3D spaces.
User’s can communicate with each other through private or public text messages or VoIP calls. Many environments are enabled with 3D positional audio which processes audio exchange between avatars according to their proximity and relative placement in the virtual space. In other words, a user can hear only those whose avatars are close enough to his and their voices fade away as avatars move away from each other. Also, voices appear to come from the left or right depending on the relative positions of avatars.
In this paper we present the results of a study we conducted to evaluate the usability and acceptability of VEs for collaboration in an enterprise. We analyse what features of a virtual environment are ∗ This work was done while Aditya Zutshi was with TCS Innovation Labs, Delhi † email:geetika.s@tcs.com
Most VEs, e.g. Second Life [Second Life ], Project Wonderland [Project Wonderland ], OpenSimulator [OpenSim ], are built on the client-server model. The server maintains all the information about the world and sends regular updates to all connected clients. The clients have a graphical rendering engine that uses the information sent by the server to render a view of the world according to the preferences, e.g. camera view etc., set by the user.
Copyright © 2009 by the Association for Computing Machinery, Inc. Permission to make digital or hard copies of part or all of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, to republish, to post on servers, or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from Permissions Dept, ACM Inc., fax +1 (212) 869-0481 or e-mail permissions@acm.org. VRCAI 2009, Yokohama, Japan, December 14 – 15, 2009. © 2009 ACM 978-1-60558-912-1/09/0012 $10.00
331
of the participants believed that the ability to change the appearance was extremely important. Most of them stated that they were uncomfortable with the interface of the VE and therefore, didn’t change their avatar’s appearance. This indicates the need for adequate training on the environment. Gestures: Most of the participants believe that avatar gestures are extremely important form of non-verbal communication. The interface for triggering a gesture in all the three clients was through keyboard/mouse. Almost all the participants found this unnatural and difficult to use. They felt the need of capturing gestures through a passive and non-invasive interface like cameras.
(a)
A large share of participants (71%) felt that a restricted set of gestures was necessary for official scenarios. They were of the opinion that situation based rules should be defined and the gestures should conform to those rules. For example, avatar’s should not be allowed to fly or dance in a meeting room. A small percentage of participants (29%) believed that in a VE, they should be free from the rules of the real world and that having situation based gestures is not necessary.
(b) Table 1: (a) Second Life and (b) RealXtend Conference Rooms
3
Camera and Graphics: A majority (57%) preferred the third person camera view to the first person view (29%). A small share of participants (13%) found both views to be equally necessary. The participants unanimously stated that good graphics were absolutely mandatory to create a feeling of immersiveness in the VE. Most of the participants (71%) felt that having a collaborative virtual space look like an office setup was extremely important.
Our Study
We conducted a pilot study of three VEs to determine their usability and acceptability for productive use in an enterprise. These were Second Life [Second Life ], RealXtend [RealXtend ] and Qwaq [Qwaq ].Out of these, RealXtend, table 1 (b), is open-source, was hosted on our own servers and had . Second Life, table 1 (a), is hosted on the Internet by its parent company Linden Labs and is free for anyone to use. Qwaq is a proprietary application whose trial version, hosted by its parent company, was used for the study. The reason for choosing environments for hosted differently was to test the performance of each hosting option.
Navigation: A large percentage of participants preferred walking to flying in the VE. They did less than 20% of their total movement by flying. A small percentage of people used flying for 20%-40% of their total movement. As mentioned earlier, a large share of the participants felt that flying should be restricted in closed spaces. Communication: Most of the participants preferred voice chat to text chat while a smaller percentage used both almost equally. Approx 85% of the participants did more than 50% of the total communication through voice chat. Around 15% of the participants used text chat slightly more than the Voice Chat. The participants feel that the ability to whisper or do side conversations is absolutely required and would make the VE feel more realistic.
The participants of the experiments were experienced professionals of the IT Industry, in the age group 25 to 55, and were familiar with the available suite of collaboration tools. Two thirds of the participants were males and less than half had experience of 3D gaming. We organized activities like general discussions, meetings, presentations, document editing and the participants were asked to fill out a questionnaire at the end. Participants were asked questions in two categories. The first consisted of questions on different features common to three VEs and the second consisted of questions on comparison of the three.
Document Editing: Most of the participants (83%) rate the feature of in-world document editing to be absolutely mandatory for collaboration. They feel that there should be features to open a document, edit, save and share in-world. There should be an easy interface supporting this feature. They also feel that it is important that others are able to see the changes someone is making. The rest (17%) find this feature to be moderately important. Approximately two-third of the participants finds the feature of desktop sharing to be absolutely mandatory for collaboration. The rest find this feature to be moderately important.
For the experiments, we used mobile workstations by HewlettPackard with Intel(R) Core (TM)2 Duo CPU T7700 @ 2.40 GHz 2.40 GHz processor, ATI FireGL Graphics Card, 2.00 GB RAM and Windows VistaTM Business Service Pack 1 to run the clients, with an Internet Connection of 1 MBPS.
3.1
General Features of Virtual Environments
3.2
User Representation: Almost all the participants (86%) were comfortable with having an avatar as a representation of themselves in the VE. The rest felt uncomfortable with the lack of similarity between the avatar and the actual person it represented in terms of facial resemblance and age. Most of the participants (86%) very strongly felt the need to have the ability to change the avatar’s appearance. The participants who were uncomfortable being represented by an avatar were among those who rated this feature extremely high. The remaining participants rated it moderately important.
Comparison between the Virtual Environments
In this section, we present the results of a comparison between the three VEs Second Life, RealXtend and Qwaq. The participants were asked to rate different features of the environment, like realism of avatars, graphics, voice clarity and so on, on a ten-point scale. Graphics: The Second Life and RealXtend virtual spaces were custom built by us while in Qwaq a meeting room was created using their templates. Second Life allows users to build their space using the graphical tool in-built in their client, while RealXtend allows this as well as meshes created in external modelling tools to
A little more than half of the participants didn’t change their appearance after logging in for the first time. This was unexpected as most
332
features for collaboration. RealXtend had the least delays and this was because it was hosted on our local network. Although Second Life and RealXtend emerged as the top two virtual worlds, they were placed in the middle of the scale. This indicates that users’ expectations from VEs for collaboration are far from being met by the current state of the art environments. (a)
(c)
(b)
4 Conclusion We found that most participants were open to using VEs for collaboration. However, the following are important for them to be effective. Firstly, as with any new technology, people need adequate training to be able to use a VE optimally. Secondly, network efficiency is extremely critical for such environments. In cases where the users are all part of one organisation, it is better to have a locally hosted environment rather than one which is external. Also, since such environments are being considered to aid and increase productivity, they must perform with minimal latency. Thirdly, realism of the environment both in terms of user representation and graphical models has a big impact on the experience. Since people know their colleagues from the real-world, they expect their avatars to also be like them and deviations can be disturbing. Also, if the virtual space is built like a real-world space the users are familiar with, their office in this case, users take to the environment more easily as they immediately know how to get around. Finally, real-world rules must apply in the VE as well. The decorum and behaviour that is observed in a real-world work environment must be followed in the virtual world as well. This may be implemented by controlling the abilities of avatars in different virtual scenarios.
(d)
Table 2: Average Score for Various Features
be uploaded. The Second Life space was built as a standard meeting room, while in RealXtend, we created a virtual replica of our own office. The Qwaq meeting room allows one to add pre-created objects like chairs, tables and screens on walls that could be whiteboards, web browsers or documents for editing. The participants rated RealXtend and Second Life quite high on the quality of graphics, table 2 (a). These environments were immersive and kept the participants engaged. Qwaq on the other hand had unrealistic graphics. RealXtend scored the highest for the perceived speed of rendering. Second Life performed almost equally well. Qwaq had a lot of delays in rendering and scored the least, table 2 (b).
References BARTLE , R. R. 2004. Designing Virtual Worlds. New Riders Publishing.
Voice Clarity and Lag: RealXtend scored the highest for voice clarity. Second Life was also rated close. Qwaq was rated very low, table 2 (c). For all the environments, one major problem faced was of echo. The experiments were conducted using the speakers and microphone in-built in the laptops. As there is no implementation of echo cancellation on the clients, a recommended practice is to use the push-to-talk mode while speaking or earphones.
B ROWN , B., AND B ELL , M. 2006. ‘there’ as a collaborative virtual environment. In Proceedings of CSCW 2006, ACM Press, ACM. E VER Q UEST . http://everquest.station.sony.com/. O PEN S IM. http://www.opensimulator.org. P ROJECT W ONDERLAND. https://lg3d-wonderland.dev.java.net/.
RealXtend had the minimum time perceived lag in audio and scored the highest, table 2 (d). Second Life also performed almost equally well. Qwaq had a lot of time lag in audio and for most of the time, the audio chat was suspended for a considerable number of participants. Also, Qwaq tried to make up for delays by speeding up the playback. This lead to uncanny and disturbing distortions in audio.
Q WAQ. http://www.qwaq.com. R EAL X TEND. http://www.realxtend.org. S ECOND L IFE . http://www.secondlife.com. T HE MUD C ONNECTOR. http://www.mudconnect.com.
Realism of Avatars: The participants rated the avatars in Second Life to be the most realistic with average score 5.71, RealXtend second best with a score of 5.16 and Qwaq third with a low score of 1.86. Although, Second Life and RealXtend were rated almost equally, their score was in the middle of the scale implying that there is much scope of improvement in avatars.
U LTIMA O NLINE . http://www.uoherald.com. W EISETH , P. E., M UNKVOLD , B. E., T VEDTE , B., AND L ARSEN , S. 2006. The wheel of collaboration tools: A typology for analysis within a holistic framework. In Proceedings of CSCW 2006, ACM Press, ACM.
Based on their experiences, the participants were asked to evaluate the different VEs taking into consideration the features and the quality of service offered by them. The participants rated Second Life the highest at 5.86 on a scale of 10. RealXtend was rated at 5.67 on a scale of 10 while Qwaq was rated at 2.57.
3.3
Discussion
Out of all the three VEs, Qwaq scored the least primarily because of unrealistic avatars, poor graphics, delays in rendering and audio and poor voice clarity even though it had the maximum number of
333
334