Contrary to the overwhelmingly positive scientific reporting on dialogue systems and speech interfaces more generally, the everyday experience of many roboticists is mostly one of brittleness and frequent failure. While some of these failures can be traced back to suboptimal speech recognition systems, other failures are of a more fundamental nature.
While humans deploy a number of repair mechanisms when misunderstandings are detected by one of the speakers - some of which involve modalities other than speech - similarly powerful repair mechanisms are lacking on the side of the speech interfaces when failures occur in human-robot interactions. The natural multi-modality of robots provides these, at least in principle, additional abilities to communicate that are unavailable to non-embodied machines and could play a fundamental role in repairing communicative failures.
This Research Topic is two-fold: Firstly, to provide an overview of commonly occurring communicative failures in human-robot interaction and, secondly, to highlight the opportunity of potential multimodal repair mechanisms to render robotic speech interfaces more resilient with respect to conversational breakdowns. We intend to fulfill the first aim by collating a variety of articles documenting and analyzing such failures with the intention to shed light onto what is largely an unreported issue experienced by many robotics practitioners. The Research Topic aims to obtain a wide, optimally representative overview of the different types of failures as part of a longer-term attempt to build a failure taxonomy. In order to achieve the second aim, we invite both articles reporting existing research on conversational repair in HRI as well as position papers outlining the potential of such mechanisms. Submitted papers may describe or propose approaches to deal with one or several types of failure categories, and we expect that the multimodal nature of robots will play an essential role in at least some of these approaches.
The Research Topic seeks to attract articles on conversational failures in human-robot interaction and potential repair mechanisms aiming to equip speech interfaces with the ability to deal with such failures. In particular, we seek articles that:
• contain in-depth analyses of conversational failures in human-robot interaction
• contain in-depth analyses of conversational failures relevant to human-robot interaction as encountered in adjacent fields such as HCI
• document attempts to categorize the different types of conversational failures
• document or propose potential repair strategies
• discuss multi-modal approaches to human-robot communication, especially in the context of failure and repair
• discuss approaches to increase the robustness of human-robot communication, especially when expecting or planning for conversational failure
• discuss the applicability of large language models (LLMs) to robotic speech interfaces and potential or actual failures incurred when deploying these
This Research Topic is linked to the workshop of the same name at the ACM Conversational User Interfaces (CUI) conference 2023 in Eindhoven, The Netherlands, from 19–21st July 2023. Contributions presented at the workshop should make an extended contribution and be expanded to include 30% original content in order to be published in this Research Topic. We also welcome contributions that were not presented at the workshop but cover relevant themes to the focus of this collection.
Contrary to the overwhelmingly positive scientific reporting on dialogue systems and speech interfaces more generally, the everyday experience of many roboticists is mostly one of brittleness and frequent failure. While some of these failures can be traced back to suboptimal speech recognition systems, other failures are of a more fundamental nature.
While humans deploy a number of repair mechanisms when misunderstandings are detected by one of the speakers - some of which involve modalities other than speech - similarly powerful repair mechanisms are lacking on the side of the speech interfaces when failures occur in human-robot interactions. The natural multi-modality of robots provides these, at least in principle, additional abilities to communicate that are unavailable to non-embodied machines and could play a fundamental role in repairing communicative failures.
This Research Topic is two-fold: Firstly, to provide an overview of commonly occurring communicative failures in human-robot interaction and, secondly, to highlight the opportunity of potential multimodal repair mechanisms to render robotic speech interfaces more resilient with respect to conversational breakdowns. We intend to fulfill the first aim by collating a variety of articles documenting and analyzing such failures with the intention to shed light onto what is largely an unreported issue experienced by many robotics practitioners. The Research Topic aims to obtain a wide, optimally representative overview of the different types of failures as part of a longer-term attempt to build a failure taxonomy. In order to achieve the second aim, we invite both articles reporting existing research on conversational repair in HRI as well as position papers outlining the potential of such mechanisms. Submitted papers may describe or propose approaches to deal with one or several types of failure categories, and we expect that the multimodal nature of robots will play an essential role in at least some of these approaches.
The Research Topic seeks to attract articles on conversational failures in human-robot interaction and potential repair mechanisms aiming to equip speech interfaces with the ability to deal with such failures. In particular, we seek articles that:
• contain in-depth analyses of conversational failures in human-robot interaction
• contain in-depth analyses of conversational failures relevant to human-robot interaction as encountered in adjacent fields such as HCI
• document attempts to categorize the different types of conversational failures
• document or propose potential repair strategies
• discuss multi-modal approaches to human-robot communication, especially in the context of failure and repair
• discuss approaches to increase the robustness of human-robot communication, especially when expecting or planning for conversational failure
• discuss the applicability of large language models (LLMs) to robotic speech interfaces and potential or actual failures incurred when deploying these
This Research Topic is linked to the workshop of the same name at the ACM Conversational User Interfaces (CUI) conference 2023 in Eindhoven, The Netherlands, from 19–21st July 2023. Contributions presented at the workshop should make an extended contribution and be expanded to include 30% original content in order to be published in this Research Topic. We also welcome contributions that were not presented at the workshop but cover relevant themes to the focus of this collection.
