Don't Worry, Just Follow Me: Prototyping and In-the-Wild Evaluation of Smart Pole Interaction Unit with Mobility

Paper Title

Don't Worry, Just Follow Me: Prototyping and In-the-Wild Evaluation of Smart Pole Interaction Unit with Mobility

Publication Info

• Topic area: Pedestrian–automated vehicle (AV) interaction in shared spaces using mobile infrastructure-based communication.
• Keywords: pedestrian-AV interaction, eHMI, smart pole, shared spaces, mobility, trust, safety, workload, field study, infrastructure-based communication.

Background and Problem

• Problem / challenge: Pedestrian-AV interactions in shared spaces are often ambiguous, with vehicle-mounted external human–machine interfaces (eHMIs) facing visibility and timing challenges. Static infrastructure lacks mobility to adapt to dynamic crossing scenarios.
• Significance: Ensuring clear, safe, and confident pedestrian-AV interactions is critical for public acceptance of AVs in mixed-traffic environments, particularly in unregulated shared spaces.
• Motivation and related work: Prior research has explored vehicle-mounted eHMIs and static infrastructure but has not addressed the limitations of visibility, occlusion, and adaptability in dynamic, multi-actor shared spaces. This study builds on these gaps by introducing a mobile, pedestrian-side communication mediator.

Solution

• Proposed approach: Smart Pole Interaction Unit (SPIU) – a mobile, pedestrian-side system equipped with LED displays and cameras to provide explicit crossing cues (e.g., "WALK," "STOP") synchronized with AV behavior.
• Novelty:
  1. Development of a mobile SPIU prototype that acts as a pedestrian-side communication mediator.
  2. Real-world, in-the-wild evaluation of SPIU effectiveness in shared spaces.
  3. Analysis of interactions between vehicle behavior, signal modalities, and SPIU mobility.
  4. Insights into the role of mobility and redundancy in pedestrian-AV communication.
• Procedure and key techniques:
  • A 2 × 2 × 2 × 2 factorial design with 16 scenarios varying CarBehavior (Stop/Go), Mobility (Moving/Stationary), eHMI (ON/OFF), and SPIU (ON/OFF).
  • Dependent variables included crossing onset time (COT), workload (NASA-TLX), trust, safety, and acceptance ratings.
  • SPIU mobility tested as stationary or moving ahead of pedestrians to guide crossings.
  • Signals synchronized with AV behavior at a fixed distance (5 m) and time-to-arrival (1.5 s).

Results

• Concrete findings:
  • Both eHMI and SPIU significantly improved trust, perceived safety, and understandability while reducing workload (NASA-TLX).
  • SPIU often yielded stronger gains than eHMI alone, with combined eHMI+SPIU showing the best outcomes.
  • Mobility effects were conditional: moving SPIU enhanced performance when signals were active but had minimal impact without signals.
• Advantage over baselines:
  • SPIU reduced hesitation and improved confidence compared to vehicle-mounted eHMI alone.
  • Combined eHMI+SPIU configuration outperformed individual signals, though with diminishing returns.
• Experiments / evaluation:
  • Field study with 21 participants in a shared-space environment without traffic lights or crosswalks.
  • Metrics included subjective ratings (trust, safety, acceptance) and objective measures (COT).
  • Results showed that Stop+signals scenarios dominated rankings, with mobility providing minor additional benefits.
• Limitations and future work:
  • Limited to low-speed, controlled environments and a small participant pool.
  • Challenges with visual occlusion and semantic clarity of signals (e.g., "STOP").
  • Future work should explore multimodal cues (e.g., audio), diverse populations, and scaling to complex, high-density scenarios.

Summary

This study introduced and evaluated a mobile Smart Pole Interaction Unit (SPIU) as a pedestrian-side communication mediator in shared spaces. The SPIU improved trust, safety, and workload compared to vehicle-mounted eHMIs, with the combination of both signals yielding the best results. Mobility effects were context-dependent, enhancing performance when signals were active. The findings emphasize the importance of explicit, redundant signals and suggest that mobile infrastructure can complement vehicle-mounted systems to improve pedestrian-AV interactions. Future work should address scalability, multimodal integration, and inclusive design for diverse environments and user groups.