next up previous
Next: Agent-based Models of emotion Up: DRAFT: Research paradigms for Previous: Will you still like

A sampler of research paradigms for lifelike synthetic agents.

In many cases one of the key requirements of useful lifelike agents is that they be engaging. This can be had through models of emotion and personality, through realistic interaction with the virtual world, through social grace, through humor, through maintenance of a relationship with the user - even over the course of many months, through wonderfully magical semantically-driven graphics, through creating a social fabric in interesting virtual worlds, and through assisting the user through lifelike collaborations. In this next section we introduce some of the research pushing intelligent agents to new hights of engagement with their users. Note that this is intended only as a sampler - the field has already grown too wide to even hope covering in a short article such as this. This is most emphatically NOT an attempt to survey all the ``important'' work in this area. Rather we seek only to pique the reader's imagination and to introduce some of the interesting problems being considered in the contemporary agents landscape.

We have loosely organized the work into four overlapping areas. The first involves inspiring agents with meaningful personalities based on rich emotional characteristics. We cover this in work associated with Botelho, Elliott, Rosseau, Sloman, and Velazquez. The second pertains to the careful molding of artificial characters that live in artificial environments. We view this in the context of work by CMU-Zoesis, PF Magic, Millenium Interactive, and Origin's Ultima Online. The third has to do with work more generally theoretical in nature that supports foundational development for the agents paradigm. We introduce this by touching on the work associated with Perlin, Badler, Cassell, Mateas, Maes, Koda, Blumberg, Doyle, Nass, and others. Lastly, we look at the role of synthetic agents in the tutoring paradigm. For this we briefly examine the work associated with Lester, Rickel, Johnson, and Andre.

In general, there is much collaboration between labs and individual researchers, and it is often true as well that there may be several projects underway at an individual site. We will attempt to refer to each of labs, researchers, and projects, as possible. Because of the great breadth of the field we encourage the interested reader to follow up, via the Web, and paper bibliographies, on each of these four types of indices for greater depth in any single area.




next up previous
Next: Agent-based Models of emotion Up: DRAFT: Research paradigms for Previous: Will you still like

Clark Elliott
Thu Dec 25 19:14:31 EST 1997