Why That Nao?

Abstract

This paper explores how humans adapt to a conventional humanoid robot. Video data of participants playing a charade game with a Nao robot were analyzed from a multimodal conversation analysis perspective. Participants soon adjust aspects of turn-design such as word selection, turn length and prosody, thereby adapting to the robot's limited perceptive abilities as they become apparent in the interaction. However, coordination of turns-at-talk remains troublesome throughout the encounter, as evidenced by overlapping turns and lengthy silences around possible turn endings. The study discusses how the robot design can be improved to support the problematic taking of turns-at-talk with humans. Two programming strategies to address the identified problems are presented: 1. to program the robot so that it will be systematically receptive at the equivalence to transition relevance places in human-human interaction, and 2. to make the robot preferably produce verbal actions that require a response in a conditional way, rather than making a response only possible.

Voice Interfaces in Everyday Life

“…However, Luger and Sellen do not focus their work on an examination of actual settings of use as they happen. More closely related to our paper here, however, is work by Pelikan and Broth [24], who inspect the interactional organisation of human-robot interactions, identifying the practical competencies through which users of the robots in question adapt their talk to improve the accuracy of spoken word transcription by the robot, demonstrating how the limitations of automatic speech recognition may be partly overcome through methodological innovation on the part of the user. Perhaps most connected to this paper is our earlier study [27], which identifies the characteristics of interaction with VUIs on mobile devices and how such interactions unfold in multiparty social settings.…”

“…Our study draws on the traditions of Ethnomethodology and Conversation Analysis (EMCA) [8,32], as is common in HCI literature (e.g. [24]), to examine the various ways in which the Echo was implicated in talk. In the main part of our study we explore the ways in which the Echo is embedded into the situational exigencies of the home (such as other activities going on during use), and how its users account for the interactional work that use involves.…”

"Do Animals Have Accents?"

“…IPAs are now readily found on many devices such as smartphones, tablets, watches, and televisions. Additionally, although such systems fail to fully mimic human talk, Pelikan and Broth [33] were able to reveal the succinctness of how people adapt their talk to an agent's needs and capabilities, making their interactions more successful. Their work focused on a dyadic face-toface conversation with a humanoid robot, and was able to reveal a number of difficulties individuals face in such talk.…”

Adapting a General-Purpose Social Robot for Paediatric Rehabilitation through In Situ Design

“…The challenges of speech recognition with social robots such as NAO, and for voice recognition with children more generally, are well documented in the HRI literature [14]. Pelikan and Broth [28], for example, note issues associated with the required turn taking between robot and human when delivering speech, which users often find difficult to adapt to. Insufficient loudness of voiced responses or unexpected statements provided by human users pose significant challenges for SARs seeking to foster natural and authentic interactions with users.…”

“…Despite design decisions to optimise the robustness of NAO's built-in speech recognition (see Section 5.5.2), verbal interaction with the SAR remained problematic. Notably, recent studies have highlighted specific issues with the NAO platform's speech recognition [28] and natural language processing with children more generally [14]. Phase 2 session observations noted frequent falsenegative responses to simple words, such as "Go."…”