A Human-like Robotic Manipulation System Implementing Human Models of Sensory-Motor Coordination

The recent advances of humanoid robotics led to achievements in the development of human-like body parts and whole bodies, as well as of perceptual systems and schemes for behavior planning and learning. For this reason, a step further can be envisaged in the application of human-like robots as experimental tools in neuroscience, for carrying on experiments that may be difficult or impossible with human beings. This paper presents a human-like robotic manipulation platform for implementing neuro-physiological models of sensory-motor coordination in human grasping, with a particular attention to its sensory system. The proposed robotic system originates from requirements imposed by neurophysiological knowledge about the corresponding human system. Hence, it is composed of sensors and actuators replicating some level of anthropomorphism, in the physical structure and/or in the functionality. Software modules implement human-like basic mechanisms of perception and learning, on which more complex architectures are developed. The system is integrated so as to be as modular as possible and to be re-arranged for validating different hypotheses.