Seamlessly Integrated Distributed Shared Virtual Environments
The huge success of the World Wide Web is a result of a number of factors. Important ones, from the point of view of this project are the distributed nature of the data, the ability for anyone to publish and for anyone to generate links to other sites. The connectivity of Web Pages grows organically by authors locally creating document names and adding new links to any existing document.
The VRML 2 standard, which was adopted in late 1997 defines a 3D description language. VRML 2 is good at describing a non-distributed (single site) 3D environment with simple animation. Such an environment is presented as a monolithic object, possibly with some delayed downloading by using inline nodes. Inter-site connections are much the same as HTML links in that an object in the environment is associated with a target URL. When the user clicks on the object, the current environment is deleted and the new one, indicated by the URL, is loaded. The user therefore sees no visual link between the two sites ? it is as if they exist in different universes.
Current proposals for allowing some inter-site visibility are restricted to connecting a zone to virtually adjacent zones. This approach requires some cooperation between the design of two zones, even between different servers, something which is contrary to Web design ethos. Zones cannot be multiply sited, which eliminates the ubiquitous inclusion that happens with the current HTML Web based documents. Arbitrary inclusion is felt to be crucial in allowing the same flexibility for distributed virtual environments as currently exists with HTML. The proposed work will investigate techniques to allow such features. It is also intended that the proposed work start from existing standards, be compatible with them and drive them forwards.
Adding a framework for interaction will give users presence in the virtual environment. Some existing schemes enable collaboration in an abstract manner, employing a rooms metaphor. These rooms abstract spaces where objects can be shared. The rooms are cloneable and an optimistic consistency scheme can be employed. In an immersive environment, the freedom available to employ such an abstract idea of a room is not immediately available. It is not clear that users will be able to manage working with cloneable 3D virtual rooms.
Network latencies cause problems with interactive environments, if they become noticeable to users. Existing network latencies are already problematical for the perceived speed of http connections. VRML does not specify a transport protocol, however most VRML models are served via http or ftp, neither of which is appropriate for an interactive environment. Whatever network protocol is employed, there will be some residual latency, and that must be addressed.
There are two extremes in controlling interaction. One requires total control of all interactions possible, and therefore has problems with extensions. The other has no control over interactions and therefore has problems with inter-site connectivity. It is expected that some middle ground will be necessary for extensible interactive growth. Some ground rules will be defined which all environments must support, however, other details can be left open and therefore change from environment to environment. The ground rules will provide a basic methodology to disguise unavoidable network latencies so that they are not noticed within the environment.