Basic Visual and Motor Agents for Increasingly Complex Behavior Generation on a Mobile Robot.

Present work addresses the guidelines that have been followed to construct basic behavioral agents for visually guided navigation within the framework of a hierarchical architecture. Visual and motor interactions are described within this generic framework that allows for an incremental development of behavior from an initial basis set. Basic locomotion agents as, Stop&Backward, Avoid, and Forward are implemented by means of fuzzy knowledge bases to deal with the uncertainty and imprecision inherent to real systems and environments. Basic visual agents as, Saccadic, Find_Contour, and Center are raised under a space-variant representation pursuing an anthropomorphic approach. We illustrate how a complex behavior results from the combination of lower level agents always connected to the basic motor agents. The proposed methodology is validated on a caterpillar mobile robot in navigation tasks directed by an object description. [ABSTRACT FROM AUTHOR]