Modeling robot cognitive activity through active mental entities

This paper aims to lay down the foundations of an approach to the development of autonomous robot control architecture based on the explicit representation of mental attitudes underlying robot behavior, considered as autonomous active entities. The approach is intended to integrate concepts from the area of distributed architectures and of mental attitude representation and it is claimed to allow the achievement of both robust and sophisticated robot behavior. Starting from an analysis of the evolution of autonomous robot control architectures, the introduction of active mental entities in the context of a distributed control architecture is discussed and motivated. Attention is then focused on two classes of mental entities, namely intention and persuasion, and their main features are illustrated and discussed. A prototypical implementation of the proposed paradigm and its application to the control of Khepera simulator are then described. A comparison with related works and a discussion of the main directions of future research conclude the paper.