Integrating Object Recognition and Action for action sentence production and comprehension in a developmental humanoid Robot

Recent models from cognitive robotics research have recently begun to appreciate the entwined relationship between language and action, and have proposed to ground robots’ language understanding in sensorimotor representations. These models, however, focused mainly on the one-way relation between language and action, neglecting its bi-directional character. Hence, the primary aim of this research is to provide an experimental and computational framework for explaining how the processes of language interact with other processes such as motor control. We intend to tackle this issue from the developmental robotics perspective. This highly interdisciplinary field of robotics benefits greatly from the direct collaboration with child psychologists and the modeling of child development phenomena in robots. One classic approach to studying child development is to look at the errors that children sometimes commit. Young children, for example, sometimes make a serious attempt to perform impossible actions on miniature objects. We suggest that these errors may result from the onsets of language and propose to test it empirically: does a child with a larger vocabulary and/or more advanced grammatical skills perform more errors than a less advanced child? The hypotheses derived from the study will be replicated in experiments with the developmental robotics platform iCub to demonstrate the impact of such an embodied cognitive approach in the design of integrated action and language capabilities in robots. This research will contribute to the knowledge of cross-talk between language and motor structures and suggest a possible developmental mechanism, and endow a robot with the ability to comprehend sentences on object interaction. The fellowship will also provide the researcher with an opportunity to further develop multidisciplinary skills in child psychology and developmental robotics methods, and transferable skills for R&D in academia and service robotics industry.
Toddlers attempting the impossible
Have you seen your toddler try to wear a hat that is obviously too small? Or ride a toy horse? These are known as 'scale errors' and were first described in 2004 by Judy Deloache and her colleagues in the United States. As an adult you are never tempted to squeeze in through the doors of a model Porsche, no matter how much you want to drive it. This is because adults easily distinguish between an ordinary object and a smaller replica. If you are interested in helping out with this study and have a child who will be between 21 and 30 months between now and September please get in touch so we can make an appointment. Plymouth Babylab is interested in how children develop, so if you have a child aged 0 to 6 years and you would like to contribute to research, please contact us at the BABYLAB on 01752 584865 or visit www.psy.plymouth.ac.uk/babylab/.