An Efficient Cognitive Architecture for Service Robots Maurizio Piaggio + DIST University of Genova, Via Opera Pia 13A, 16145 Genova, Italy piaggio@dist. unige. it Renato Zaccaria* DIST University of Genova, Via Opera Pia 13A, 16145 Genova, Italy renato@dist. unige. it ABSTRACT This paper faces the problem of devising an efficient cognitive architecture for an autonomous mobile robot. The reference scenario is that of a service robot, able to move among people in civilian environments, for carrying out assistive tasks. Examples are night surveillance tasks in large buildings like museums, exhibitions, banks and warehouses; transporting things; escorting people in departments, and exhibitions. A common characteristic of these situations is an intrinsic trade-off between complexity and predictability. From one point of view, in fact, the environment may change in time and space; from the other side, a large amount of relevant situations can be classified a priori, as well as the most part of the topological and geometrical characteristics of the environment. In this paper, a cognitive model and its underlying software architecture for such robots are presented, focusing the attention on the integration of the different necessary cognitive components: + tel: +39 010 353 2801; fax: +39 10 353 2154 *tel: +39 010 353 2798; fax: +39 10 353 2154 177