|Thesis abstract: |
Robots nowadays are not widespread among human beings but are still confined into plants and research laboratories; this is mainly because they lack of smart autonomous behaviors and they are not able to learn and adapt in response to changes both in the environment and their physical structure; they still miss what is called cognitive development. A new view in robotics and artificial intelligence which states that these cognitive processes are strongly tied to the physical structure of the body and its interaction with the environment is now emerging; this new theory is in contrast with the functionalism and dualism of cognitivist and classical AI and tries to highlight the importance of embodiment and interaction for learning in cognitive systems .
This innovative point of view in IA affects different aspects of cognitive processes, ranging for example from language and symbolic reasoning, to motor learning and grasping for interacting with the environment; in  is described an embodied system that follows a developmental route that goes initially through the exploration of its body and terminates into the characterization of external objects by grasping. A second important project concerning cognitive and embodied systems is the iCub platform   which mainly aims at testing the paradigm which says that manipulation plays a fundamental role in the development of cognitive capability.
This research wants to explore artificial intelligence and robotics from this new perspective; by means of embodied cognitive systems we want to deal with the problem of cognitive development and learning, focusing in particular on those aspects related to movements and to the grasping task, which play a fundamental role for cognition in human beings. There are various aspect that have to be integrated and among these is worthwhile to mention symbolic reasoning, imitation learning and neural coding, that is the way our brain stores concepts and information .
Aim of this research is to implement a cognitive bio-inspired architecture able to show child-like basic learning mechanism in terms of high level symbols emergence and movements imitation and generation for object grasping with regard to specific tasks. Some tests are to be done to validate our architecture both in simulation and on real humanoid robot, the NAO robot.
 W. J. Free e R. Núñez, «Restoring to cognition the forgotten primacy of action, intention and emotion,» Journal of consciousness studies, 1999.
 L. Natale, F. Orabona, G. Metta e G. Sandini, Sensorimotor coordination in a "baby" robot: learning about object thorugh grasping, Genova.
 G. Metta, G. Sandini, D. Vernon, L. Natale e F. Nori, «The iCub humanoid robot: an open platform for research in embodied cognition,» in PerMIS'08, Gaithersburg, USA, 2008.
 G. Sandini, G. Metta e D. Vernon, «The iCub cognitive humanoid robot: an open-system research platform for enactive cognition,» 50 Years of AI, pp. 358-369, 2007.
 B. B. Averbeck, P. E. Latham e A. Pouget, «Neural correlations, population coding and computation,» Nature, vol. 7, pp. 358-366, 2006.
Keywords: learning, imitation, symbolic, movement, grasping, embodied, interaction, humanoid robotics, cognitive systems