Does a “robot dog” need legs, ears, and tail? A comparative analysis of intention- and emotion-attribution to Miro-E and Unitree Go1

The study reported in this paper analysed the effectiveness and acceptability of ethologically inspired expressive behaviours implemented in two distinctively different embodiments of the zoomorphic robots Miro-E and Unitree Go1. It investigated how primary school children attribute intentions and emotions to the two robots, examining the importance of certain body-parts in human-robot interactions to convey affective states and express intentions (e.g. ears, tail, legs). 111 students aged 7 to 10 years participated in the study in a within-subject design, observing an interaction between each robot and an experimenter in small groups. Every child observed both robots interacting with an experimenter in the same scenario following an AB-BA order. After each interaction, a questionnaire was presented to each student individually. Effects of a) robot embodiment, b) dog-ownership, and c) students’ age on their perception of the robots, focusing on differences between the two robots’ emotionally and intentionally expressive behaviour, were analysed. Results identified significant effects of each independent variable. While the Miro-E robot was identified as expressing emotions better - underlying the importance of affective features such as ears, and a tail - there was no significant difference in children’s intention-attribution to the two robots, and Unitree Go1 was selected as the preferred one over Miro-E. Despite the differences both Miro-E and Unitree Go1 reliably conveyed the intended emotions and intentions, providing further evidence for the applicability of the ethorobotics approach. Findings implied that the incorporation of zoomorphic embodiment features to express social signals could expand potential applications of these robots.