Communicating with Avatars in Virtual 3D Worlds
Volker Thoma
Andrea Haf
Arno Hitzges
Fraunhofer Institut für Arbeitswirtschaft und Organisation
Nobelstr. 12, D 70569 Stuttgart Germany
+44 0711 970 2341
volker.thoma/andrea.haf/arno.hitzges@iao.fhg.de
Keywords
Virtual communities, usability testing, avatars, multi user worlds, chat, real time communication, metaphors
Extended ABSTRACT
Virtual community environments (VEs) provide shared experience of VRML worlds and HTML pages as well as communication between their visitors [6]. The main goal of the usability study presented here was to evaluate current user interfaces and interaction concepts with the focus on ease of access and ease of use. Furthermore, the degree of newcomers’ acceptance for this technology was explored. Most Internet users do not yet exploit the possibilities to interact online with other persons using virtual environments that offer not only chat functions but complete worlds with rooms, places and objects [5]. Virtual communities provide a perfect platform for cooperative work, distributed learning, and for new forms of online shopping and trading [4]. Thus, virtual environments offer new and exciting opportunities not only for experienced web users [3]. However, current virtual environments exhibit rather complex user interfaces. To explore the potential of these multi-user environments for usage by customers in virtual shopping malls we studied computer literate users who had no prior experience with VE, just like newcomers to online services.
There are yet only a few applications which offer the possibilities of 3D multi-user environments such as avatars (i.e. a user’s virtual representation), robots and shared events. In our usability tests, we compared two popular yet in some ways different concepts of 3D multi-user worlds: Blaxxun’s community (community B) and Active Worlds of Ring of Fire Studios Inc. (community A) which are shown in figures 1 and 2 and briefly described in Table 1.
The main components of current 3D multi user worlds are:
- chat functionalities
- handling of avatars and gestures
- graphical user interface controls
The usability of these components was evaluated for the two applications with special emphasis of control and feedback given. A further research questions concerned the acceptance of robot avatars and the assistance they offer to users?
Figure 1: Blaxxun’s Ccpro. The thumbnails to the right represent hyperlinks to rooms, in the lower center the chat fields are displayed, at the lower right the tab control shows lists of worlds, avatars, options etc.
Figure 2: Active World. The tab control to the left displays a list of worlds, the lower center comprises the chat functions. At the top of the screen are the menu and task bars. In the VRML window (center) a user is represented as a human figure (avatar).
| Active World (A) |
Blaxxun (B) |
|
| CHAT |
Text chat |
Text chat |
Recipients |
Public |
Public/private |
| Feedback |
speech bubbles (VRML window), chat protocol |
chat protocol |
| GESTURES |
Actions |
Actions, verbal descriptions (chat) |
| Control |
Task bar |
Pull down
menu, |
| Feedback |
VRML window |
VRML window,
|
| Context |
World |
Avatar |
| GUI |
Application |
Internet Browser |
| Control |
Menu/task
bar, |
Tab control,
links, |
| AVATARS |
User defined according to world context |
User defined |
| Selection |
Avatar list |
Avatar Gallery |
| User list |
No |
Yes |
Table 1: Characteristics of the 3D worlds in the tested configurations. (Note that Active X was a demo version)
Usability Tests
Ten users aged 11 to 69 with academic and non-academic backgrounds took part in the usability tests. The users’ tasks included approaching an avatar (controlled by a "wizard" evaluator), talking to an avatar (chat), using gestures (waving to an avatar), navigating through the world via mouse and/or keyboard, changing one’s own avatar and finally changing to another world. After completion of the tasks short interviews were held with the participants.
Chat Functions
In both applications most users did not recognize the chat input field at the bottom of the screen, even after extensive search and help by the instructor. When users could not find the chat input field they searched for it in the tab controls, in the symbol bar and in the menu bar. Some users thought they had to enter their chat text in the VRML window. Many users were tempted to click on an avatar to initiate a chat. When this did not work they clicked on the avatar’s name above the avatar. Clearly users expected most actions to take place in the VRML window.
In Blaxxun‘s CCpro many users expected the chat option under a button with a face icon that opens a drop down list with gestures. Since the gestures were offered in the form of complete sentences the users suspected that chat is confined to these predefined sentences. In the public chat protocol, not only the chat texts but also the gestures of other avatars appeared in the form of sentences ("X smiles at Y"). However, users didn’t notice the difference between gestures and chat texts of other avatars, although they were presented in different colors.
Often users did not realize when they were talked to. Most users were so entrenched in navigation tasks that they hardly recognized when the "wizard" evaluator tried to chat with them. During the tests users were asked to chat with the avatar "Andrea". However, users were not certain whether the text they had entered was actually delivered to the target avatar. Also, the participants were irritated by the texts displayed above the avatars in the VRML window (A). These texts contained the name of the avatar and its users last input (chat). The text lines were often hard to read, in particular when several avatars were in the scene at the same time, which caused overlapping texts.
In summary, even when participants became accustomed to the standard chat interface they found it difficult and strenuous to use and could not keep track of the conversations in the protocol field. Clearly, there was a need for a private chat option, but participants in Blaxxun’s community did hardly find or remember the option in the tab control under the list of avatars. Rather, people click directly on the avatar they wish to interact with.
Avatar Gestures
Most users felt the need to see their own avatar, in particular at the beginning of the test session. Five users thought that another avatar which they saw in the VRML window was their own, some of them had this impression several times. When users of Active Worlds were asked to initiate an action or gesture (dance, wave) via the task bar they were uncertain about the effect of their selection as long as they remained in the default 1st person view. Just a few users thought of checking whether their avatar actually performed an action by switching to the 3rd person view.
In Blaxxun’s CCpro few users discovered the gesture menu that was hidden behind a button with a smiling face icon. Eight of the users searched in the tab control for initiating a gesture. Users also searched for labeled buttons to initiate or select gestures. Some users thought that the gestures had to be entered as text into the chat line instead of using the drop down menu. Users were not sure whether they successfully initiated a gesture and they often did not recognize gestures of other avatars.
In general, information and options referring to action and interaction in the VRML world were not suspected in the menus and tab controls. It was hard to see which tab was active. Furthermore, the right mouse button was rarely used in Blaxxun’s World, and thus the related context menus were found accidentally, if at all. Users had difficulties to select options from the cascading menus and complained about the long lists of items. Ambivalent symbols, wording and color coding were found in both applications and often prevented users to complete their tasks. Virtual environments offer a variety of options and functions that refer to very different tasks such as chat, navigation, and control of the interface. There was no coherent concept for the user where to look for functions or actions across platforms. Users were constantly searching for options, even when they had used them before.
Interviews and Questionnaires
In interviews, most users accepted robot avatars that offer assistance. However, participants wanted robots to be reserved and offer help in the form of short dialogues and only when help is needed.
In a preference questionnaire Blaxxun was clearly preferred in respect to "Overview of avatars" because of its list that indicates the number of avatars that are inside a world. Moreover, the possibilities of changing an avatar by choosing one in a "gallery" were preferred in B. However, users clearly preferred A when it came to the question about "ease of initiating gestures". The possibility to initiate gestures by simply activating a push button is one reason. Another reason is the way emotions are phrased in B (e.g.:"XY is happy"). Users were confused whether these sentences referred to chat functions. No substantial differences of preference ratings were found in terms of learnability, appeal of graphic design, and general ease of use (Figure 3).
Figure 3: Number of users that preferred application A or B. Indifferent responses are not shown.
Recommendations for Design
Virtual 3D worlds offer new opportunities for online communities, but as for now lack of usability shuts out many potential participants. Users had a very hard time chatting and navigating in the 3D worlds although they liked it, as interviews and acceptance ratings reveal. The major design flaw are breaks in the metaphor of a 3D world. Although the users should feel immersed in a 3D world, they always are forced to switch back to a 2D user interface with tab controls, menu bars and cascading menus. Therefore, we propose to put more information and interaction into the VRML window:
Reduce 2D GUI functionalities, Use World Context
Use cascading menu lists, tab controls, etc. sparingly. Try to integrate menus in the world (e.g. into virtual billboards). Use information in the 3D world to extend the world metaphor. For example, a "private chat" could be held in a private room, "public" chats in public places. Employ direct manipulation (click on objects, avatars).
Integrate Chat into the World
Show names above avatar, but not complete chat texts. Show only the texts of people next to or in front of the user's avatar, and avoid overlapping text fields. A prompt in input fields is recommended. A click on an avatar should initiate a – private or public - chat.
Provide Feedback
Users missed feedback about their avatar's appearance and actions (gestures). For example, a "mirror" tool could be used in accordance with the metaphor.
Make Selecting Actions Natural
Gestures and actions should not be presented as sentences. Use icons (e.g. faces) and only short text description if possible. Use indicative wording rather than complete sentences. Make sure gesture control and chat functions are separated. Users should be able to select gestures like avatars from a gallery. A preview of gestures and actions should be possible.
User ratings indicate a high degree of acceptance of 3D worlds despite major usability problems. However, most users were likely to abandon the 3D worlds because important functionalities such as chat and interaction could not be understood and handled. The user tests presented here indicate that for a many potential users the point of indignation is reached very soon when using current VE interfaces. Future developments should aim at integrating functions and information into the 3D world by utilizing and extending the "world" metaphor. This will attract a broader user group and allows the exploitation of the commercial and social potentials of virtual environments.
REFERENCES:
- Blaxxun interactive AG http://www.blaxxun.com, 1998
- Circle of Fire Studios Inc http://www.activeworlds.com ,1998 5
- Hagel, J., Armstrong, A.G.: Net Gain. Harvard Business School Press, Boston, 1997.
- Jupiter communications, New Media Report (8/98), New York, 1998 2
- Mynatt, E. D.; Adler, A.; O´Day, V. L.: Design for Network Communities, in CHI´97 proceedings (1997), ACM Press 1
- McGrath, A.; Oldroyd, A.: Designing Shared Virtual Environments, in CHI´98 Proceedings (1998), ACM Press
