Affordances as a framework for robot control

The concept of affordances, with its empha-sis to the interactions between the robot and the environment, is highly relevant for epige-netic robotics. However, the use of the con-cept in robotics has been rather rudimentary, mostly confined to an inspiration source. In this paper, we present a new formalization of the concept, based partially on the recent for-malizations proposed in Ecological Psychology and Linguistics, which provide a framework for robot control, learning and planning. We argue that affordances, as relations within the robot-environment system, can be seen from three dif-ferent perspectives; namely agent, observer and environmental. We argue that affordance rela-tions can be learned from the interactions of the robot within its environment. The formalism ar-gues that the interactions of the robot, can be represented as a nested triple of the form (effect, (entity, behavior)) indicating that the behavior applied in an environment perceived as the en-tity, would produce a perceivable effect. It is sug-gested that these nested triples of raw sensory-motor data obtained from different interactions can be used to form four different equivalence classes towards the formation of affordance re-lations. We present three studies implement-ing certain aspects of the formalism on a mo-bile robot moving in an environment filled with different types of objects. Specifically, we show that, (1) the formation of the entity equivalence classes corresponds to the perceptual learning of affordances, (2) the formation of effect equiva-lence classes, followed by the formation of entity equivalence classes can lead to the development of goal-directed behaviors from a set of primi-tive ones, and (3) the formed equivalence classes and relations provide support for planning and deliberation. 1