The status of representations in behavior based robotic systems: the problem and a solution

Abstract

We deal with the status of representations in behavior based robotic systems (BBRS). We present a brief description of the role of representation in AI since its inception. Then we discuss a tentative taxonomy of various approaches, after the behaviouristic turn in the field, according to their treatment of representation. Within the interactionist approach we propose a theory of agency and interaction. In the sequel, we propose an experimental setup based on this theory, and present simulation results in order to illustrate the notion of intrinsic representations. In the concluding section we situate our approach in the space of BBRS, discuss its advantages and shortcomings, and outline ideas for future work.