Robot Self-Assessment and Expression: A Learning Framework

Abstract

Future autonomous robots that operate in teams with humans should have capabilities that facilitate intuitive and/or implicit communication, for example in the form of emotional expressions. These emotional expressions should be presented clearly to the human to promote adequate understanding of robot behaviors and intent. In this paper, we present a Robot Self-Assessment and Expression framework derived from reinforcement theory of motivation and the current state-of-the-art in machine learning. The proposed framework theoretically describes how a robot could display emotional expressions depending on both predicted outcomes and actual outcomes of a task. The end goal for this framework design will be for the robot to obtain anticipatory guidance and performance feedback from a human instructor during a training task. Future research and areas for testing and validation of the framework are discussed.