A relation-based model for animating adaptive behavior in dynamic environments

Over the past few decades many advances have been made in computer animation, which has largely replaced traditional hand-drawn animation. However, the traditional path model used in animation has not changed. This model as based on predefined paths does not effectively address the needs of scene animation, which demands unpredictable environmental influences, multiple motion interactions, and personal behaviors. Some of the suggested approaches to scene animation are sensor-effector model, rule-based model, and predefined environment model, but these approaches don´t completely solve the problem. This paper presents a relation-based model for dealing with the important issues of scene animation. Two fundamental principles of our model are: the specification of atomic units of motion (relations), and the ability to combine these atomic units to produce sophisticated behaviors. Each relation models a simple responsive behavior between two objects, and has its own sensor, response, duration, strength, and state. The dynamic states of relations can be interactively structured into hierarchical layers that produce complex scene behaviors. During the animation, structured relations are dynamically triggered by either its sensing condition(s) or the state of the environment. The variable state-control hierarchies facilitate on-line scene behavior animations. Several dancing examples for the use of our model are illustrated