A New Based-on-Artificial-Intelligence Framework for Behavioral Animation of Virtual Actors Andres Iglesias Dep. of Applied Mathematics, University of Cantabria, SPAIN iglesias@unican.es Francisco Luengo Department of Computer Science, University of Zulia, VENEZUELA fluengo@cant.net Abstract The realistic animation of the behavior of virtual actors emulating human beings has been a very dynamic field of research during the last few years. A major goal is to create a behavioral system for the virtual actors so that they behave as realistic as possible. Among the large number of different techniques to solve this problem, those based on Artificial Intelligence (AI) do represent a powerful (but not well explored yet) approach. In this paper, we present a new framework for behavioral animation of virtual actors. The framework applies several Artificial Intelligence techniques (neural networks, expert systems, fuzzy logic, K-means) to build a sophisticated behavioral system so that the actors can take intelligent decisions by themselves. The paper describes the general framework, its main components and how these AI techniques have been effectively applied to this purpose. Some programming issues, the main steps of the simulation flow and some illustrative examples are also analyzed in this paper. 1. Introduction The realistic simulation and animation of the behavior of virtual actors emulating human beings has attracted much attention during the last few years [1- 8,12-17]. Most of this interest has been motivated by its remarkable applications to the entertainment industry, specifically in movies and video games. However, the range of potential applications also includes Architecture, Science, advertising and others. A major challenge in behavioral animation is the construction of a framework able to integrate the different techniques required for the realistic simulation of virtual humans. Among them, we can include those for perception, motion control, 3D rendering and animation, goal selection, action execution, communication between actors, their interaction with the environment, etc. The goal is to provide the virtual actors with a high degree of autonomy, so that they can evolve freely, with a minimal input from the animator. In addition, this animation is expected to be realistic; in other words, the virtual actors must behave according to reality from the point of view of a human observer. A number of different approaches have been proposed to fulfill the previous requirements [4,6- 7,9,12-17]. However, while the quality of pictures has been dramatically improved during the recent years, relatively little effort has been placed upon the simulation of the human cognitive processes (learning, memory, recognition, etc.) By contrast, this is the primary goal of the Artificial Intelligence (AI) field. Because of that, some recent papers have proposed the application of AI techniques to this purpose [7,9]. In this paper we propose a unified framework that provides all features mentioned above. On one hand, it allows animators to create a virtual world in which virtual actors evolve autonomously in a very realistic way. On the other hand, the framework incorporates several AI techniques so that the virtual actors can perform actions on the basis of intelligent decisions. The structure of this paper is as follows: in Section 2 we describe the architecture of the framework and some programming issues. The simulation flow is described in detail in Section 3. Section 4 describes briefly some illustrative examples available at Internet. The main conclusions of this work close the paper. 2. Framework architecture 2.1. Graphical, physical and behavioral systems Due to its inherent complexity, it is convenient to decompose the proposed framework into simpler components, which can be rather assigned to one of the following sytems (see Figure 1): Proceedings of the International Conference on Computer Graphics, Imaging and Visualization (CGIV’04) 0-7695-2178-9/04 $20.00 © 2004 IEEE