[論文レビュー] Robot-Wearable Conversation Hand-off for Navigation

Dániel Szabó, Aku Visuri+2 more|arXiv (Cornell University)|Feb 16, 2026

Social Robot Interaction and HRI被引用数 0

ひとこと要約

要約: 本研究は stationary な社会ロボットと wearable watch 之间の cross-device conversation hand-off を用いた室内ナビゲーション支援の実装と評価を行い、興味深いと評価されつつもパフォーマンス向上は見られず、ウェアラブルのみのナビゲーションを高く好むことが示された。

ABSTRACT

Navigating large and complex indoor environments, such as universities, airports, and hospitals, can be cognitively demanding and requires attention and effort. While mobile applications provide convenient navigation support, they occupy the user's hands and visual attention, limiting natural interaction. In this paper, we explore conversation hand-off as a method for multi-device indoor navigation, where a Conversational Agent (CA) transitions seamlessly from a stationary social robot to a wearable device. We evaluated robot-only, wearable-only, and robot-to-wearable hand-off in a university campus setting using a within-subjects design with N=24 participants. We find that conversation hand-off is experienced as engaging, even though no performance benefits were observed, and most preferred using the wearable-only system. Our findings suggest that the design of such re-embodied assistants should maintain a shared voice and state across embodiments. We demonstrate how conversational hand-offs can bridge cognitive and physical transitions, enriching human interaction with embodied AI.

研究の動機と目的

• Investigate whether a conversation hand-off from a stationary robot to a wearable device can support indoor navigation.
• Compare user experience and navigation performance across robot-only, wearable-only, and robot-to-wearable hand-off conditions in a university campus setting.
• Identify design considerations for multi-embodiment AI conversation hand-offs in public spaces.
• Demonstrate a concrete implementation where a shared voice/state enables seamless hand-off across embodiments.

提案手法

• Implement a cross-device conversational navigation system using a Pepper robot and an Apple Watch Ultra 2.
• Use a local server to coordinate speech-to-speech pipeline with Mimic 3 for synthesis and Whisper for recognition.
• Employ Rasa for dialogue management with constrained, task-focused interactions (no open-domain chat).
• Design a deliberate, explicit hand-off trigger with confirmations to maintain transparency between devices.
• Evaluate three conditions (Robot-only, Wearable-only, Robot-to-Wearable Hand-off) in a within-subjects study with N=24 on university-campus routes.

実験結果

リサーチクエスチョン

• RQ1 RQ1: How does hand-off of a human-AI conversation from a stationary embodiment to a wearable embodiment affect user experience?
• RQ2 RQ2: How does the hand-off affect navigation performance?
• RQ3 RQ3: What design considerations can be identified for AI conversation hand-offs between wearable and robot embodiments?

主な発見

Conditions / Routes	Metric	Robot	Wearable	Hand-off
Robot	Mean Task Time (sec)	140.8	157.6	192.5
Robot	Task Time SD (sec)	57.9	79.6	88.2
Robot	Error Rate (%)	62.5%	50.0%	50.0%
Robot	Mean Interactions	2.9	3.7	4.2
Robot	Interaction SD	1.9	1.2	2.1
Routes	Mean Task Time (sec)	173.9	145.8	171.2
Routes	Task Time SD (sec)	79.8	84.1	70.5
Routes	Error Rate (%)	12.5%	75.0%	75.0%
Routes	Mean Interactions	3.3	3.4	4.0
Routes	Interaction SD	2.1	1.6	1.7

• Participants preferred the wearable-only system (70.8%), with no one preferring robot-only navigation.
• Robot condition showed higher mental demand and effort compared to Wearable and Hand-off, but performance remained high across conditions.
• Trust in the system did not differ significantly across conditions, though Wearable and Hand-off elicited more neutral/slightly positive responses on certain questions.
• There were no statistically significant differences in task completion time or error rates across conditions, though Hand-off produced slightly higher mean interactions.
• The hand-off was generally perceived positively when smooth, and participants valued having access to instructions via the watch during movement.
• Qualitative feedback highlighted benefits of portability, privacy, and continuity of guidance across devices, as well as some usability improvements (e.g., push-to-talk vs. hotword, verbosely detailed directions).