Power Chess : robot-to-robot nonverbal emotional expression applied to competitive play

Human-machine communication has evolved from one-to-one to multi-agent systems where the interplay between machines themselves interacts with human perception and behavior, complicated by unconstrained emotion-based variables in social systems. To investigate Human-Robot and Robot-Robot-Human interaction while constraining the interaction variables in a rule-based system, we developed an artistic intervention using competitive game performance between robotic arms. Two robots play chess with each other while expressively making gestures like thinking, examining, hesitating, shows of satisfaction and bewilderment, breathing, etc. These nonverbal behaviors and evolving rules between games tell a narrative of power struggle between two robots of aggressive vs. reflective personalities. We used recorded videos to assay audience interpretations of individual and robot-to-robot expressions, finding that gestures like standing and confirming were perceived as aggressive, while head turns, deliberation, and audience alerts were seen as curious. Human perception of robot play-style and their own intended play strategies were influenced by robot-robot interactions, such as holding defensive strategies when the robot was deemed aggressive. Robotic movements caused audiences to attribute personality characteristics to them, modifying their intended strategy in patterns like pretending to be friendly first to lull the robot opponent. Our work uses artistic metaphors to study multi-agent environments that cannot be easily controlled for in scientific settings.