Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Facial Features Integration in Last Mile Delivery Robots

Delgermaa Gankhuyag, Stephanie Groiß, Lena Schwamberger, Özge Talay, Cristina Olaverri-Monreal

TL;DR

This study addresses pedestrian acceptance of last-mile delivery robots by introducing expressive facial features on an LMD robot and measuring initial reactions. It conducts a campus-based experiment comparing Happy, Angry, and Neutral facial expressions against a baseline, analyzing data with Chi-square, Cramér's V, and a binary logistic regression to relate facial expressions to observed emotion. Findings indicate that expressive faces elicit more responses than no expressions, while neutral expressions significantly reduce happy reactions, yielding moderate predictive accuracy (AUC ≈ 0.66). The work suggests that simple facial cues can enhance interaction with LMD robots and potentially improve acceptance, though generalizability is limited and future work should broaden emotions and participant demographics, and explore adaptive facial expressions.

Abstract

Delivery services have undergone technological advancements, with robots now directly delivering packages to recipients. While these robots are designed for efficient functionality, they have not been specifically designed for interactions with humans. Building on the premise that incorporating human-like characteristics into a robot has the potential to positively impact technology acceptance, this study explores human reactions to a robot characterized with facial expressions. The findings indicate a correlation between anthropomorphic features and the observed responses.

Facial Features Integration in Last Mile Delivery Robots

TL;DR

This study addresses pedestrian acceptance of last-mile delivery robots by introducing expressive facial features on an LMD robot and measuring initial reactions. It conducts a campus-based experiment comparing Happy, Angry, and Neutral facial expressions against a baseline, analyzing data with Chi-square, Cramér's V, and a binary logistic regression to relate facial expressions to observed emotion. Findings indicate that expressive faces elicit more responses than no expressions, while neutral expressions significantly reduce happy reactions, yielding moderate predictive accuracy (AUC ≈ 0.66). The work suggests that simple facial cues can enhance interaction with LMD robots and potentially improve acceptance, though generalizability is limited and future work should broaden emotions and participant demographics, and explore adaptive facial expressions.

Abstract

Delivery services have undergone technological advancements, with robots now directly delivering packages to recipients. While these robots are designed for efficient functionality, they have not been specifically designed for interactions with humans. Building on the premise that incorporating human-like characteristics into a robot has the potential to positively impact technology acceptance, this study explores human reactions to a robot characterized with facial expressions. The findings indicate a correlation between anthropomorphic features and the observed responses.
Paper Structure (13 sections, 2 equations, 8 figures, 4 tables)

This paper contains 13 sections, 2 equations, 8 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Background Information
  3. Experimental Setup
  4. Apparatus
  5. Conducted Experiment
  6. Variables selection, model design and analyzing methods
  7. Response variable
  8. Covariates
  9. Model design
  10. Results
  11. Results of Chi-square and Cramer's V tests
  12. Model's result
  13. Conclusion and Future Work

Figures (8)

  • Figure 1: LMD robot with the happy face.
  • Figure 2: LMD robot with the angry face.
  • Figure 3: Adapted figure representing the Johannes Kepler University Campus jku
  • Figure 4: Facial expressions design sheet utilized to select the optimal facial feature for the conducted experiment darkspeeds
  • Figure 5: Sample sizes of covariates by count
  • ...and 3 more figures