Abstract

Inside this developing research field of robotics there is a trend that looks at building bio-mimetic robots. Bio-mimetic humanoid robots must possess the ability to emulate human beings physically and “mentally” . These robots must be clever, i.e. they must be able to learn and use the acquired cognitions in the different steps of their life. Machine learning is one of the developing fields of anthropomorphic robotics. Learning robots are able to learn from their actions thanks to a trial and error training phase and they are able to act using the acquired cognitions, replying consistently to problems already resolved in the training phase. The attempt to build robots able to learn from their actions comes from the temptation to understand the biological working that let human beings able to do it. The main and still far goal of this robotic field is the realization of intelligent robots with autonomy of decision and action, human-like independent operators able to interact with human beings and the environment, learning from their actions. In this paper we present our project, which is called FACE . It is a work in progress that provides parallel steps. It is not limited to specific goals but it tries to open different fields of discussion for future uses in several fields. FACE is a learning robot able to learn from the environment in which it “lives” and nowadays is used as behavioural aid for children with autism. Tomorrow, going beyond, FACE will be an intelligent robot with interactive presence and with emergent innovative behaviours that comes from the interaction with the environment