Using a Declarative Language to Describe the Interactions in Virtual Scenes

A generic platform useful to design and run virtual scenes as a result of a human-like description is introduced. The present article describes the operation of one of the main components of such platform: the virtual-environment (VE) editor. The main goal of the VE editor is to allow a human description of a virtual scene and translate it, if errors-free, into a set of low-level commands. The modeler is provided a declarative language useful to: a) arrange the virtual entities using geometric constraints instead of geometrical data, b) assign personalities to virtual humans, and c) specify goals to be fulfilled by virtual humans. A scene takes place in a specific world. For every different world to model a set of rules must be defined. The VE editor allows the specification of some features related to a particular virtual world, including the following: actions and skills that can be performed, virtual entities acceptable, pre-and post- conditions for interactions, effect of natural laws (gravity, collisions), and a set of rules, which reject or allow the assignment of goals according to the state of virtual humans