Impacts of Design and Behavior on the Acceptance of Anthropomorphic Service Robots: Statistical Modeling of Post-Experiment Questionnaire Responses

Acceptance of robotics in customer service is complex as modern robots have brought about increasingly natural dialogue with users. Prior research has established significant relationships with respect to usefulness and ease of use in modeling acceptance of robots. However, a systematic understanding of the causal relationships between the design and behavior of robots and their acceptance is lacking. In this paper, we use questionnaire responses from 146 participants as part of a publicly available dataset in affective computing in HRI to analyze acceptance of service robots. After validating the framework used in the study, we observe significant relationships with respect to anthropomorphism, transparency, liability, and morality. We find no differences in acceptance based on anthropomorphism. We find that neither transparent communication nor failures resulting in liability issues cause significant changes in acceptance. We find that overtly immoral behavior by robots is more harmful to acceptance than moral behavior is beneficial. Our results are surprising, and crucial for understanding user perceptions of robots based on design and behavioral factors.