Toward Learning Interactive Behavior

This document is divided into two parts. The first part details the authors opinion on the three questions posed by the ICRA NEWHRI 2008 workshop. The second part represents a brief introduction to the research conducted by the author toward realization of autonomous robots that can learn interactive behavior through real world interactions with humans. I. CHARACTERIZING HRI A. What characteristics unify and should be common to all areas of HRI research? HRI research is by definitions an interdisciplinary research area. It is about humans as much as it is about robots, but more importantly it is the interaction between these two different kinds of creatures that gives rise to the questions and challenges addressed by HRI. So what is common to all areas of research in HRI is the concept of interaction as a dynamical relation between humans and robots. This unifying concept is there in every kind of research in this domain including the analysis, design and implementation techniques that can produce a more effective interaction in the domains of interest. HRI areas of research share more than just a common concept, they share a strong real-world orientation and human centered methodology. While most subfields of robotics are real world oriented, this is even more important in HRI because of the existence of the complex intentional and sometimes unpredictable human in the loop. HRI is by definition a human centered endeavor because the human is not only the final consumer of the solution HRI is producing but more importantly he/she is a part of challenges this field is addressing. B. What properties typify novel and significant research in