The importance of faces in human-robot interaction

  • Neelu Gurung

    Student thesis: Master's Thesis


    This thesis investigates an interactive robotic art installation by the artist Stelarc in the context of Human-Robot Interaction. In particular this thesis examines how robot faces, movement, and speech help or hinder the interactive experience between the robot and humans. To address this problem, firstly I studied the impact of speaking with audio on the impression formation of appearances. Secondly, with the aid of an industrial robot arm with and without a face, I investigated the effect of three different embodied faces (alive face, static face, and no face) on subjective and objective variables when interacting with a robot. Lastly, I explored if similar traits exist in human-human interaction about safety that can be observed in human-robot interaction concerning the equilibrium point. In the first experiment, I used different appearances of the artist Stelarc, which included real photographs, low fidelity digital renderings, and high-fidelity avatars of the artist. I measured how speaking and non-speaking affects forming an impression of the face. In the second experiment, I used subjective and objective measures to explore the proxemic behaviour of humans in the vicinity of interactive robots. Not surprisingly, I found that human-like voices emanating from a robot positively affect a human’s impression formation, while machine-like voices negatively impact impression formation. Surprisingly, I found several other measures thought to be crucial in human-human impression formations to be inert in the human-robot context. An important finding was the significant change in perceived safety in robots held by the participants before and after an interaction. The main contributions of the thesis include insights into the role of faces in HRI, particularly the counter intuitive finding that faces may not be as important as once thought to be for a successful human-robot interaction. Evidence was also established as to improving the perceived safety of robots in close interactive settings, such as with collaborative robots. And new insights were gained into the Uncanny Valley hypothesis with implications for avatar design
    Date of Award2022
    Original languageEnglish
    SupervisorDamith Herath (Supervisor), Shuangzhe Liu (Supervisor), Janie Busby Grant (Supervisor) & Sarbazhosseini (Supervisor)

    Cite this