Exploiting analogical representations

The paper proposes to use analogical instead of symbolic representations and to annotate these representations dynamically with markers and structures that emerge by the parallel local activity of simple computational agents.

Abstract

The paper discusses the use of internal models for autonomous agents. It proposes to use analogical instead of symbolic representations and to annotate these representations dynamically with markers and structures that emerge by the parallel local activity of simple computational agents. Two experiments illustrate the ideas of the paper: one in hand-eye coordination and one in the acquisition of a world map by a wandering mobile robot.