HCC-SGER: Social and Moral Interaction Patterns with a Personified Robot

University Of Washington, Seattle WA

Investigators

Abstract

A genre of robots has begun to move from the pages of science fiction to the research laboratories, and even into society at large. These robots, in various ways and to varying degrees, have a persona, are adaptive and autonomous, and can communicate, learn, use natural cues, and self organize. Such "personified" robots, which will become part of people's everyday lives, form the focus of this research wherein a new approach to their design will be explored that builds on a concept the PI terms "interaction patterns." By an interaction pattern the PI means characterizations of essential features of social interaction between humans and robots, specified abstractly enough such that many different instantiations of the pattern can be uniquely realized given different types of robots, purposes, and contexts of use. A simple example of an interaction pattern between humans takes place when we are introduced to a new person. In Western society, we often shake hands, say "hi," exchange our names, and perhaps engage in a little chit-chat about the weather. In other cultures, people might bow to one another, or offer a namaste greeting. We can call this universal social activity an "Introduction" and its purpose is to facilitate further social interaction. While the Introduction is never enacted exactly the same way twice, the activity is structured and follows a recognizable pattern. The PI argues that this pattern is but one of many that can be used to help structure human-robot interaction. In the current project, the PI will generate between 50-75 social and moral interaction patterns. He will specify a methodology for designing interaction patterns (which, in part, draws from characterizing successful patterns of interaction in human-human interaction and human-nature interaction), will articulate the optimal levels of characterization between abstraction and concretization for interaction patterns, will show multiple instantiations of individual patterns (to highlight that patterns are not rigid molds but engender a multitude of varied instantiations), will implement technically a sequence of these patterns with an actual robot, either autonomously or (more often) through a "wizard-of-oz? technique, will establish a beginning set of guidelines for validating interaction patterns, and finally will begin to validate a handful of sequenced patterns. The technical implementation and beginning validation of the patterns will occur using the humanoid robot Robovie from Advanced Telecommunications Research (ATR) in Japan. Broader Impacts: If the PI's approach proves successful, and is then taken up by other laboratories, it will reshape the existing landscape in the human-robot interaction (HRI) research community. Furthermore, it is likely in the near future that the public will raise serious concerns about the introduction of personified robots into modern society; this research will provide a proactive design agenda that recognizes the legitimate societal concerns and shows how we can "do it right" with robots.