guest ::
TY - THES AU - Dautzenberg, Pia Sophie Charlotte TI - Social navigation in HRI : experimental investigation of human-robot-following scenarios PB - Rheinisch-Westfälische Technische Hochschule Aachen VL - Dissertation CY - Aachen M1 - RWTH-2024-06815 SP - 1 Online-Ressource : Illustrationen PY - 2024 N1 - Veröffentlicht auf dem Publikationsserver der RWTH Aachen University N1 - Dissertation, Rheinisch-Westfälische Technische Hochschule Aachen, 2024 AB - Service robots that take over the transport of goods for their users are thought to be a promising solution for increasing urban traffic challenges. The publicly funded project ‘UrbANT’ aimed to develop such a service robot. A central feature of UrbANT (robot named after the project) is that it follows its user, which poses special demands on the HRI design. It must be designed in such a way that it (1) supports users to successfully navigate together with the robot and (2) encourages intention to use. Existing literature often emphasizes the positive effect of anthropomorphic design. Human-like robots seem more likeable, competent, and socially warm. Moreover, interaction with a human-like entity is more intuitive. But do these findings also apply to service robots like UrbANT? How do human-like, machinelike, and animal-like design aspects of communication and appearance affect evaluation and - particularly important for this type of robot - spatial interaction? These questions were investigated in four studies in which communication and appearance of UrbANT were varied systematically. In study 1, an online study, nine different communication styles were examined while UrbANT’s appearance was kept constant (boxy body, six wheels, rocker-bogie chassis, display in the front). In addition to human-likeness and machine-likeness, aspects such as gender of voice, formality of language and modality of communication were varied. A key finding of this study is that the human-like communication styles were rated more competent, warmer, and less discomfortable than the machine-like ones. Differences between female and male voices, on the other hand, could not be found. Study 2, a laboratory study, aimed to replicate the results of study 1 in direct contact with the robot. Furthermore, the study investigated whether there are interactions between communication design (human-like vs. machinelike) and preferred following distance of the robot during joint navigation. In fact, some findings of study 1 could be replicated. Regarding following distance, it was found that with both communication variants a shorter following distance was desired as it conveys a stronger feeling of control. Study 3, again an online study, varied the display design and thus appearance of the robot. Participants were presented with two human-like, two animal-like and two machinelike design concepts. In contrast to study 1 and study 2, hardly any differences could be found between participants evaluations of the design variants. A key conclusion of this study was that appearance may be less important than communication in assessing parameters such as competence, warmth, and discomfort. In study 4, a second laboratory study, appearance and communication were combined. In addition to examining subjective evaluation, this study investigated whether the two design concepts (human-like vs. machinelike) lead to differences in participants navigation behavior when mastering a course together with the robot. It showed that participants not only evaluated UrbANT more competent, warmer, and less discomfortable when UrbANT exhibited human-like communication and appearance features but, furthermore, tended to navigate more considerately. Altogether, this dissertation and its studies contribute important insights and recommendations for the design of HRI in the context of service robots that follow their users. Overall, it appears that service robots like UrbANT benefit from anthropomorphic characteristics, especially when communication and appearance features are combined. Nevertheless, the mix of functional appearance and human characteristics seems to be particularly beneficial. Besides contributing insights into HRI design this thesis sheds light on the comparability of results gathered in online and laboratory studies. LB - PUB:(DE-HGF)11 DO - DOI:10.18154/RWTH-2024-06815 UR - https://publications.rwth-aachen.de/record/989510 ER -