Multi-Modal eHMIs: The Relative Impact of Light and Sound in AV-Pedestrian Interaction

Title of the Paper

Multi-Modal eHMIs: The Relative Impact of Light and Sound in AV-Pedestrian Interaction

Paper Information

• Subject Area: Interaction design between autonomous vehicles and pedestrians
• Keywords: Autonomous vehicles, External Human-Machine Interface (eHMI), Vulnerable Road Users (VRU), pedestrians, vehicle-pedestrian interaction, multimodal interface

Research Background and Issues

• Identified Problems or Challenges:

  • Autonomous vehicles lack traditional driver communication methods (e.g., eye contact, gestures), which may make it difficult for pedestrians to judge the vehicle's intentions.
  • Current technologies are primarily visual-centric, overlooking the potential of auditory or multimodal communication.
  • Solely visual communication may be less effective in scenarios with obstacles, distracted pedestrians, or limited visibility.

• Significance:

  • Enhancing interaction efficiency between autonomous vehicles and pedestrians is critical for road safety.
  • Exploring the potential of multimodal eHMIs to improve interaction efficiency and user experience contributes to designing safer and more human-centered autonomous vehicles.

• Research Motivation and Related Work:

  • Existing studies largely focus on the design of visual eHMIs, with limited research on auditory or audio-visual multimodal interfaces.
  • Sounds (e.g., chimes or low-frequency tones) have been shown to attract pedestrian attention in certain contexts, but their potential when combined with visual signals has not been systematically studied.

Proposed Solution

• Proposed Methods or Solutions:

  • Design three types of eHMIs: slow-pulsing light strips (visual), chimes (discrete auditory signals), and humming sounds (continuous auditory signals).
  • Systematically compare the effects of these eHMIs (single visual, single auditory, and combined visual-auditory) on pedestrian interaction.

• Innovations:

  • The first experimental study to investigate the effectiveness of multimodal eHMIs in conveying the yielding intentions of autonomous vehicles.
  • Proposed insights into how different modality combinations influence pedestrian crossing decisions and subjective user experience.

• Implementation Steps and Key Techniques:

  • Conduct video-based experiments where participants watch videos of approaching autonomous vehicles and record their crossing intentions in real-time.
  • Test the impact of different eHMI modes (unimodal and multimodal) on pedestrian crossing intentions.
  • Use the User Experience Questionnaire (UEQ) to evaluate the user experience of the eHMIs.
  • Collect participants' subjective rankings and discussion feedback on multimodal eHMIs.

Research Findings

• Specific Results:

  • Objective Data:
    • The presence of eHMIs significantly improves pedestrians' ability to understand the vehicle's yielding intentions.
    • No significant differences in objective impact were observed between different modalities (visual vs auditory vs combined).
    • Multimodal combinations (e.g., light + chime + hum) did not significantly increase crossing intentions, though some participants found them helpful for confirming vehicle intentions.
  • Subjective Data:
    • Users showed a preference for the combination of chimes and light signals, while expressing dissatisfaction with the full multimodal combination (light + chime + hum), citing information overload.

• Comparison with Existing Solutions and Advantages:

  • Provides a systematic comparison of unimodal and multimodal eHMIs, addressing the research gap regarding auditory signals in vehicle communication.
  • Highlights the trade-offs between the complexity of multimodal designs and user experience.

• Limitations and Future Directions:

  • The experiment was conducted in a laboratory setting, lacking validation in real-world dynamic traffic environments.
  • Cultural differences in understanding eHMIs were not considered.
  • Future research should focus on complex environments involving multiple vehicles and pedestrians, as well as large-scale scalability issues.

Based on the above findings, the study indicates that while multimodal eHMIs do not outperform unimodal ones in objective performance, they hold significant potential for enhancing user experience and meeting accessibility needs. It is recommended that future designs carefully consider user preferences and environmental complexity factors.