Interfacing intelligent software to robotic peripherals

Abstract

A paradigm for the specification of robotic behavior is proposed, based on a template of control and data elements replicated in a hierarchical lattice. Considerations from physiological psychology are proposed as control elements for the model. The resulting model allows freedom in implementational details within a symmetrical design, and should also provide a framework for evolving robotic behavior from relatively primitive to extremely complex. The model's implications for realizing learning capabilities in robotic systems are discussed, and its implementation and demonstration are described.