A global workspace theory model for trust estimation in human-robot interaction

Abstract

Successful and genuine social connections between humans are based on trust, even more when the people involved have to collaborate to reach a shared goal. With the advent of new findings and technologies in the field of robotics, it appears that this same key factor that regulates relationships between humans also applies with the same importance to human-robot interactions (HRI). Previous studies have proven the usefulness of a robot able to estimate the trustworthiness of its human collaborators and in this position paper we discuss a method to extend an existing state-of-the-art trust model with considerations based on social cues such as emotions. The proposed model follows the Global Workspace Theory (GWT) principles to build a novel system able to combine multiple specialised expert systems to determine whether the partner can be considered trustworthy or not. Positive results would demonstrate the usefulness of using constructive biases to enhance the teaming skills of social robots.