When a Robot Communicates Through Air: Contextual Interpretations of Wind and Olfactory Cues

Paper Title

When a Robot Communicates Through Air: Contextual Interpretations of Wind and Olfactory Cues

Publication Info

• Topic area: Human–Robot Interaction (HRI) using nonverbal sensory cues.
• Keywords: Human–Robot Interaction, olfactory cues, wind cues, nonverbal communication, contextual framing, ambient signals, affective interaction, Wizard-of-Oz study, sensory modalities.

Background and Problem

• Problem / challenge: Nonverbal, sensory communication channels like olfactory and wind cues remain underexplored in HRI, particularly in mobile robots. Existing research focuses predominantly on visual and auditory cues, which often demand directed attention and may disrupt users during tasks.
• Significance: Understanding air-mediated cues could enable robots to deliver subtle, ambient notifications that are less intrusive and more contextually appropriate, enhancing their integration into everyday environments.
• Motivation and related work: Prior studies have investigated wind and scent as functional or ambient feedback but have not explored their use as intentional social signals in mobile robots. This paper addresses the gap by examining how these cues are perceived across different behavioral contexts.

Solution

• Proposed approach: Development and evaluation of a mobile robot system, ScentBot, which delivers olfactory and wind cues to communicate implicit social intent.
• Novelty:
  1. Empirical investigation of olfactory and wind cues as communicative channels in HRI across multiple contexts.
  2. Creation of a data-driven framework to characterize user interpretations of these cues.
  3. Expansion of the HRI design space by highlighting the potential of air-mediated cues for subtle and affective communication.
• Procedure and key techniques:
  • Conducted a Wizard-of-Oz study with 35 participants across six contextual scenarios (e.g., over-working, proper-resting).
  • Delivered three stimuli types (scent with weak wind, weak wind, strong wind) using ScentBot, a mobile robot equipped with a scent dispenser and airflow system.
  • Used quantitative measures (Likert scales) and qualitative thematic analysis to evaluate participants’ perceptions of the robot’s intent and emotional impact.

Results

• Concrete findings:
  • Scent cues were rated highest for care and positive impressions, while strong wind was perceived as directive but unpleasant.
  • Weak wind was often interpreted ambiguously, depending on context.
  • Behavioral misalignment (e.g., over-working) heightened the perceived relevance of cues, whereas proper-working/resting contexts reduced their impact.
• Advantage over baselines:
  • Scent cues elicited more positive affective responses compared to wind cues.
  • Strong wind was effective for directive communication but risked being perceived as coercive.
• Experiments / evaluation:
  • 18 trials per participant in randomized order, combining six situational contexts and three stimulus types.
  • Quantitative analysis using Cumulative Link Mixed Models (CLMM) and qualitative thematic coding of open-ended responses.
• Limitations and future work:
  • Potential confounds from the robot’s facial display and scent lingering effects.
  • Limited generalizability due to specific contexts (office and lounge) and short-term exposure.
  • Future work should explore long-term deployment, adaptive cue delivery, and culturally diverse interpretations.

Summary

This study demonstrates that air-mediated cues—scent and wind—can function as implicit social communication channels in HRI, shaping users’ perceptions of a robot’s intent and social role. Scent was consistently associated with care and positive affect, while strong wind conveyed directive intent but risked being perceived as intrusive. Weak wind showed contextual flexibility but was often ambiguous. These findings highlight the importance of modality-context congruence in designing subtle robotic interventions. Scent is recommended for supportive engagement, strong wind for behavioral correction, and weak wind for gentle nudges. Future research should address limitations such as olfactory habituation and explore multimodal combinations for adaptive, long-term use.