<p>Breathing is a fundamental biological rhythm linked to cognition, attention, and social perception. While robots often incorporate anthropomorphic features to enhance engagement, the effect of breathing-like behaviours on the perception of robots and on human-robot interaction (HRI) is largely unexplored. We examined whether simulated robotic breathing influences engagement towards a humanoid robot and its perception. We tested 75 participants (30 men, median age 33) interacting with Softbank Robotics’ Pepper under breathing (simulated respiratory movements, breathing-like sounds) and control (Perlin noise-based movements, purposeless sounds) conditions in a crossover randomised design. Outcome measures included gaze time (primary outcome), duration of human response to robot’s questions, total human-robot interaction duration, ratio of gaze time to interaction duration, Godspeed questionnaire, and in-house questions. Participants spent significantly more time looking at the breathing robot (+ 10.65s [95% confidence interval 2.878; 18.419]), <i>p</i> = 0.0084), with longer interactions <b>(</b>+ 27.68s [12.521; 42.848], <i>p</i> = 0.0006), but shorter responses (-8s [-15.596; -0.414], <i>p</i> = 0.0390). Breathing also led to higher perceived animacy (+ 0.69 [0.106; 1.282], <i>p</i> = 0.0224) and intelligence (+ 0.53 [0.001; 1.061], <i>p</i> = 0.0490). A multivariate model identified robot’s breathing and two participants traits (empathy and attitude toward robots) as independently associated with gaze time. Gaze time did not depend on humans’ awareness of robot’s breathing. We conclude that the effects of robotic simulated respiration align with the primordial embodying and enlivening power of breathing. The underlying cognitive mechanisms remain to elucidate, but this indicates that simulated breathing could meaningfully enhance social robotics, assistive AI, and virtual agents.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Breathing Life Into Robots: Pseudo-Respiration Enhances Human-Robot Interaction—a Behavioural Study

  • S. Lavault,
  • C. Desmons,
  • D. Hajage,
  • A. Mazel,
  • F. Plumet,
  • P. Gauthier,
  • C. Peiffer,
  • M. Chetouani,
  • M-C. Niérat,
  • C. Pelachaud,
  • M. Khamassi,
  • Thomas Similowski

摘要

Breathing is a fundamental biological rhythm linked to cognition, attention, and social perception. While robots often incorporate anthropomorphic features to enhance engagement, the effect of breathing-like behaviours on the perception of robots and on human-robot interaction (HRI) is largely unexplored. We examined whether simulated robotic breathing influences engagement towards a humanoid robot and its perception. We tested 75 participants (30 men, median age 33) interacting with Softbank Robotics’ Pepper under breathing (simulated respiratory movements, breathing-like sounds) and control (Perlin noise-based movements, purposeless sounds) conditions in a crossover randomised design. Outcome measures included gaze time (primary outcome), duration of human response to robot’s questions, total human-robot interaction duration, ratio of gaze time to interaction duration, Godspeed questionnaire, and in-house questions. Participants spent significantly more time looking at the breathing robot (+ 10.65s [95% confidence interval 2.878; 18.419]), p = 0.0084), with longer interactions (+ 27.68s [12.521; 42.848], p = 0.0006), but shorter responses (-8s [-15.596; -0.414], p = 0.0390). Breathing also led to higher perceived animacy (+ 0.69 [0.106; 1.282], p = 0.0224) and intelligence (+ 0.53 [0.001; 1.061], p = 0.0490). A multivariate model identified robot’s breathing and two participants traits (empathy and attitude toward robots) as independently associated with gaze time. Gaze time did not depend on humans’ awareness of robot’s breathing. We conclude that the effects of robotic simulated respiration align with the primordial embodying and enlivening power of breathing. The underlying cognitive mechanisms remain to elucidate, but this indicates that simulated breathing could meaningfully enhance social robotics, assistive AI, and virtual agents.