An adaptive approach for dynamically generating behavior nets on intelligent agents

The behavior-based approach has recently emerged. Its success is a result of the bottom-up development of coupled control processes, which are specifically designed for a particular situation. However, this approach has some limitations owing to flexibility and its lack of goal-directedness. We propose a new approach to the modeling of a behavior nets systems, which is a core component of an agent´s control architecture. The model, its main features, its most important implementation issues and the results of some experiments carried out in order to produce new behavior patterns and global behavior are presented