THE PROBLEM TO BE ADDRESSED AND ITS RELATION TO THE STATE OF THE ART:
Virtual reality is one of the most exciting areas of development in computing today. Systems which completely immerse the user in the virtual world, known as immersive systems, allow the most realistic experience of virtual environments developed thus far. An aspect of immersive systems that holds a great deal of promise involves the use of graphical icons, called avatars, representing real users in the virtual world. This project aims to investigate communication in a virtual world using avatars. Applications will be developed to address different aspects of communication such as verbal communication (with which we are most familiar), communication using physical gestures, and communication using facial expressions (e.g. of mental/emotional states). Attention will also be paid to inter-avatar interaction with respect to the manipulation of objects and use of tools, especially co-operatively.

SURVEY OF RESOURCES AND LITERATURE:
There is a considerable quantity of literature relevant to this project. Various web-sites offer information on virtual reality and university sites similar to the Rhodes Virtual Reality Special Interest Group appear promising. A particularly interesting site is Hannes Högni Vilhjálmsson's web site, which details his "Avatar Interaction" MSc. thesis, and various articles on modes of interaction for virtual environments and related topics.

Previous research has been conducted at Rhodes on using avatars, specifically Luis Casenueva's MSc. thesis entitled "Minimal Motion Capture with Inverse Kinematics for Articulated Human Figure Animation". Although not dealing with exactly the same issues as this project, Luis' project contains certain relevant aspects, as do Matt Mundell's Honours project "Exploring Immersive Virtual Interface Design using a Simple VR Image Viewer" and Mike Rorke's MSc. thesis "Virtual Reality Interaction Techniques". Another past thesis that may prove useful in this project is Gail Shaw's "Hand gestures as a method of interacting with Virtual Reality" as this deals with one of the avenues under consideration for selecting and choosing communication options such as facial expressions. Psychological literature dealing with communications, especially non-verbal communications, may be useful. There are a host of texts on the subject, both on the web and in psychology journals, books and research publications.

DESIGN CONSIDERATIONS AND IMPLEMENTATION DECISIONS:
The design for the project should allow for the development of a richly populated virtual environment. The avatars themselves should be designed so as to allow human-like physical gestures, expressions and movements. Research has been done on facial modeling by tracking points on the face with a view to accurately rendering facial expressions. This has proved to be quite a complex procedure involving a great deal of processing. An alternative approach possibly more suitable for avatar communication, would involve designing a set of pre-computed facial expressions for each avatar representing particular mental states or emotions e.g. happiness or sadness. Digitized images could also be developed, for particular users' personal expressions. These facial expressions could be chosen by the user from a set menu and used to enhance the quality of the interaction, allowing the non-verbal cues we rely on in real-life situations. It would be beneficial to design a generic expression menu and allow for an easy way to associate actual images with this, so as to allow pre-computed images to be exchanged for digitized images if the user prefers this option. Verbal communication could be received via a microphone, and simply digitized and sent either just to its intended recipient, or to all avatars participating in the virtual environment. An interesting idea still under investigation is the development of speech synthesis capabilities, allowing textual input from a keyboard. This would allow objects/entities to communicate directly with users, and would allow remote users and those without immersive technologies to participate in the virtual environment. A suggestion has been made to investigate the issue of privacy in this arena of avatar communications, for example allowing privileges to be granted to each user determining who can hear what and when. This would enable private conversations to take place "in public". There are various interesting possibilities for implementing this, ranging from simply leaving those not involved in a conversation "out of the loop", to scrambling the audio so that users without the correct privileges to hear a specific conversation would hear something that sounded like another language.

The interaction between a number of avatars and one or more objects/tasks will be also investigated. A joint activity such as a tug-of-war or an extension of previous research using a ball passed back and forth (e.g. ping pong games etc) would provide a good starting point for this part of the study. Physical gestures will also be investigated as a means of communication, including a study of naturalistic vs. fantastical gesture representation and users responses to these. Polhemus trackers can be used to map user movements into an avatar equivalent in the virtual world; this has been implemented successfully in the past. The representation of the other avatars in the users' visual field is an important consideration. The response of a sample of users to various avatars and styles of communication will be noted in order to determine the most effective, intuitive and popular elements.

The project will be implemented on the CoRgi platform already in existence at Rhodes University, in order to take advantage of previous research and applications. It will be coded in the form of a C/C++ application according to object-oriented principles, in order to allow easy extension of existing base classes such as the VRActor class. Once a basic version of the project has been developed there is potential to investigate avatar communication within a distributed virtual world, by allowing communication over a low-bandwidth network

THE EXPECTED RESULT:
Following further research and careful design and experimentation, the project aims to have identified and developed effective methods of communication for use with avatars in a virtual environment. These methods should facilitate interaction between avatars, allowing rich and varied communications of a diverse nature, and should hopefully contribute to the enjoyment and the immersive quality of the virtual world. The project has potential applications in distance education, virtual conferencing, virtual chat-rooms and in counseling situations with children and other such cases in which the user is likely to interact better with an avatar than with another person in real life.

POSSIBLE EXTENSIONS:
1. With appropriate software, specifically accurate free speech voice recognition software, the system could be extended to allow a textual representation of speech in the virtual world, for example in the form of speech bubbles.
2. Successful communication between avatars will allow the development of immersive interactive games.
3. The system could be extended to allow communication over a low-bandwidth network, which would facilitate a richly interactive video-conferencing system.