Knowledgeable Talking Robots Luigia Carlucci Aiello, Emanuele Bastianelli, Luca Iocchi, Daniele Nardi, Vittorio Perera, and Gabriele Randelli Department of Computer, Control, and Management Engineering Sapienza University of Rome Via Ariosto 25, 00185 Roma, Italy lastname @dis.uniroma1.it Abstract. Speech technologies nowadays available on mobile devices show an increased performance both in terms of the language that they are able to capture and in terms of reliability. The availability of perfor- mant speech recognition engines suggests the deployment of vocal inter- faces also in consumer robots. In this paper, we report on our current work, by specifically focussing on the difficulties that arise in grounding the user’s utterances in the environment where the robot is operating. Keywords: Speech Processing, Natural Language Understanding, Knowledge Representation, Robotics. 1 Introduction Natural language interaction with computers and other artifacts is a long stand- ing goal. The latest technologies in speech understanding now available on cheap computing devices or through cloud computing have been successfully employed in a variety of consumer products, such as mobile phones or car navigation sys- tems. Such systems not only allow the user to control specific functionalities of these devices by means of vocal commands, but also allow for more general forms of interaction, such as for example, querying the internet. As robots are being marketed for consumer applications, viz. telepresence, cleaning and en- tertainment, vocal interaction is expected to be supported to make them more appealing and accessible to the user. The first consumer robots with vocal interfaces are currently being released in the market and significant improvements towards talking robots are expected in the near future. The implementation of natural language interfaces for robots brings a number of technological challenges emerging from the engineering of state of the art technologies into practical solutions suited for complex devices such as robots. Besides some complexities related to the control of action capa- bilities, the interaction with talking robots is particularly challenging because of the relationships of the robot behaviors with the operational environment. As an example, consider the command “Go to the kitchen:” in order to be able to execute it the robot must know at least where the kitchen is. Moreover, if the command includes a request such as “to check whether there is coffee,” it may be desirable that the robot also knows what kind of objects can be found in a kitchen. K.-U. K¨ uhnberger, S. Rudolph, and P. Wang (Eds.): AGI 2013, LNAI 7999, pp. 182–191, 2013. c  Springer-Verlag Berlin Heidelberg 2013