A Comparison of Descriptive and Emotive Labels to Explain Human Perception of Gait Styles on a Compass Walker in Variable Contexts

Autor: Wali Rizvi, Umer Huzaifa, Amy LaViers, Jacey Lambert
Rok vydání: 2019
Předmět:
Zdroj: RO-MAN
DOI: 10.1109/ro-man46459.2019.8956361
Popis: The behavior and aesthetics of robots can impact perception by human viewers, and prior work has shown that context influences this judgement. This paper presents an experiment to better understand what sort of gait label can best explain human estimate of an internal state based on external changes, despite effects of variable context on the perception of gait. The study analyzes how a user’s perception of movement in a simple two degree-of-freedom mechanism changes through the use of varying environments via more emotive or more descriptive labels. Specifically, five bipedal gaits were overlaid onto illustrated backgrounds that were created to reflect various affective inclinations and given labels with and without emotive implications. Users were then asked to rate the accuracy of the descriptive or emotive labels of these videos, and the differences between their ratings were compared throughout the multiple backgrounds. It was found that while both sets of labels scored well, emotive labels were slightly preferred overall. However, although the addition of an environment positively affected the user perceptions in rating the suggested descriptive labels, it was more likely to negatively affect the ratings of emotive labels. The results of this analysis suggest that lay end-users prefer to make judgements about motion behavior in an emotive space but that descriptive labels may be more stable identifiers across various contexts for robot designers. These results highlight the emotional connection that humans make with motion and the role the context plays in helping to create this experience.
Databáze: OpenAIRE