Much research, both in the theoretical as in the practical, remains to be pursued as this project continues to evolve. Part of the next phase of our research will focus on expanding the functionality of the agent architecture and improving the flexibility of persistence. In terms of the space and interaction, our goal is to develop an interface that will offer natural modes of interaction. The thought of replacing the wand with a `magic mirror', for improving the direct manipulation of objects and adding to the fantasy, is currently being discussed.
Presently, the only artifact created, besides the obvious video documentation of the interaction, is saving the scene into a VRML object. We would like to expand the idea of the artifact into something that does not require a computer for it to be experienced. Our next extension towards this direction, involves the actual written output of the story. Every action can be printed out in written form using simple phrases that describe it, much like a text-based MUD. Snapshots of the agent models may be included as images in the printout, which the child can then keep as a record of the interaction.
Finally, we would like to define a methodology for qualitative and quantitative assessment of learning in the environment. Significant user testing remains to be performed in order to evaluate the efficacy of the system and justify its purpose as a learning environment. The process of incorporating virtual reality into the school environment is highly complex and involves substantial financial and human resource investment. Thus, conducting educational assessment is a significant point that needs to be addressed.