HumanoidTurk: Expanding VR Haptics with Humanoids for Driving Simulations

Paper Title

HumanoidTurk: Expanding VR Haptics with Humanoids for Driving Simulations

Publication Info

• Topic area: Use of humanoid robots as haptic feedback systems in VR driving simulations.
• Keywords: VR haptics, humanoid robots, motion feedback, immersion, driving simulation, user study, simulation sickness, fidelity, adaptability, versatility.

Background and Problem

• Problem / challenge: Existing VR haptic systems are often single-purpose, limited in fidelity, and either bulky (e.g., motion platforms) or localized (e.g., wearable devices). Humanoids have not been explored as versatile haptic media for immersive experiences.
• Significance: Enhancing immersion and realism in VR through scalable, versatile, and embodied haptic feedback could transform applications in gaming, training, and rehabilitation.
• Motivation and related work: Prior work has focused on haptic feedback via motion platforms, wearables, and robotic systems, but these approaches lack adaptability and versatility. Humanoids, with their human-like form and multi-DOF manipulation, offer potential for repurposing as haptic feedback providers.

Solution

• Proposed approach: HumanoidTurk, a system that repurposes humanoid robots to deliver whole-body haptic feedback by translating in-game g-force signals into synchronized chair motions in VR driving simulations.
• Novelty:
  1. Repurposing general-purpose humanoids as embodied haptic feedback systems.
  2. Empirical evidence demonstrating improved immersion and realism compared to conventional feedback methods.
  3. Identification of fidelity, adaptability, and versatility as key themes for humanoid-mediated haptics.
• Procedure and key techniques:
  • G-force signals from a driving simulator are filtered and mapped to humanoid arm motions.
  • Two synthesis methods (filter-based and threshold-based) were tested, with the filter-based approach selected for smoother feedback.
  • A two-phase user study evaluated the system against no-feedback, controller vibration, and human-delivered motion feedback.

Results

• Concrete findings:
  • Humanoid feedback significantly improved immersion (p < .01), realism (p < .01), and enjoyment (p < .001) compared to no-feedback and controller vibration.
  • Simulation sickness scores were highest for the humanoid condition, with nausea (M = 17.89), oculomotor discomfort (M = 18.95), and total SSQ score (M = 20.57).
  • Latency of the system was measured at M = 34.7 ms, SD = 3.4, within acceptable real-time bounds.
• Advantage over baselines:
  • Humanoid feedback outperformed no-feedback and controller vibration in immersion, realism, and enjoyment.
  • Compared to human-delivered feedback, humanoid feedback was more consistent but less adaptive.
• Experiments / evaluation:
  • 16 participants experienced four conditions: no-feedback, controller vibration, humanoid+controller, and human+controller.
  • Metrics included the Simulation Sickness Questionnaire (SSQ), User Experience Questionnaire–Short (UEQ-S), and custom immersion, realism, comfort, enjoyment, and suitability ratings.
  • Interviews highlighted trade-offs between realism and comfort.
• Limitations and future work:
  • Limited to short driving sessions and a small participant sample (n=16).
  • Robot overheating and operational noise constrained evaluations.
  • Future work should explore longer sessions, diverse tasks, and comparisons with commercial motion platforms.

Summary

HumanoidTurk demonstrates the potential of humanoid robots as versatile haptic feedback systems in VR driving simulations. By translating in-game g-force signals into synchronized chair motions, the system significantly enhances immersion, realism, and enjoyment, albeit with trade-offs in comfort and simulation sickness. The study identifies fidelity, adaptability, and versatility as key themes for humanoid-mediated haptics, positioning humanoids as a promising modality for immersive experiences. Future research should address hardware limitations, expand use cases, and explore adaptive feedback mechanisms.