Value-based human–robot-interaction. A perceptual control theory approach toward socially intelligent agents

Abstract

Recent developments in artificial intelligence boost the interaction possibilities with social robots. Since verbal communication with robots no longer needs to be scripted, it becomes even more important to guide such interactions through a value frame. So far, research has called for the consideration of values in the design phase of social robots. In addition to the value-driven design logic for social robot development during built-time of robots and human–robot interaction, we call for the dynamic consideration of values during the actual interaction with robots during run-time. We take a perceptual control theory perspective and suggest a conceptual framework for value representation as a foundation for value-based interactions of social robots. The key feature of this approach is that social robots shall be able to dynamically decide during an interaction how to best adhere to which human values instead of statically displaying pre-programmed behavior that is aligned with predefined values. We discuss the implications of this framework for implementing social robots in organizations and outline avenues of future research.