[Paper Review] HumanoidTurk: Expanding VR Haptics with Humanoids for Driving Simulations

DaeHo Lee, Ryo Suzuki+1 more|arXiv (Cornell University)|Jan 26, 2026

Social Robot Interaction and HRI0 citations

TL;DR

The paper repurposes a humanoid robot as a whole‑body haptic medium for VR driving, mapping in‑game g‑forces to synchronized chair movements and comparing smoothed versus threshold synthesis methods across user studies.

ABSTRACT

We explore how humanoid robots can be repurposed as haptic media, extending beyond their conventional role as social, assistive, collaborative agents. To illustrate this approach, we implemented HumanoidTurk, taking a first step toward a humanoid-based haptic system that translates in-game g-force signals into synchronized motion feedback in VR driving. A pilot study involving six participants compared two synthesis methods, leading us to adopt a filter-based approach for smoother and more realistic feedback. A subsequent study with sixteen participants evaluated four conditions: no-feedback, controller, humanoid+controller, and human+controller. Results showed that humanoid feedback enhanced immersion, realism, and enjoyment, while introducing moderate costs in terms of comfort and simulation sickness. Interviews further highlighted the robot's consistency and predictability in contrast to the adaptability of human feedback. From these findings, we identify fidelity, adaptability, and versatility as emerging themes, positioning humanoids as a distinct haptic modality for immersive VR.

Motivation & Objective

• Motivate and demonstrate how humanoid robots can serve as versatile haptic media beyond social/assistive roles.
• Develop HumanoidTurk to translate driving simulator g-forces into synchronized, whole-body feedback.
• Evaluate two synthesis methods (smoothed vs threshold) and compare against no feedback, controller vibration, and human feedback.
• Assess effects on immersion, realism, comfort, enjoyment, and simulator sickness.
• Identify themes of fidelity, adaptability, and versatility in humanoid-mediated haptics.

Proposed method

• Use Unitree G1 humanoid with RH56DFTP hands to grasp a chair and deliver motion feedback synchronized to driving g-forces.
• Map real-time in-game g-forces from Assetto Corsa via SharedMemory to chair motion using inverse kinematics of the humanoid.
• Compare two synthesis pipelines: (i) Smoothed filter-based mapping (low-pass for slow drift, high-pass/jerk for fast changes) and (ii) Threshold-based event-driven mapping.
• Implement safety constraints (displacements, rotations, max weight) and a safety stop mechanism.
• Conduct pilot studies to choose the preferred method, then run a main user study with four feedback conditions (No-Feedback, Controller, Humanoid+Controller, Human+Controller).
• Measure latency (total ~34.7 ms, IK-dominant) to ensure real-time feedback.

Experimental results

Research questions

• RQ1 Does humanoid-mediated haptic feedback enhance immersion, realism, and enjoyment in VR driving compared to controller-only feedback or no feedback?
• RQ2 How do humanoid-based feedback and human-delivered feedback compare in terms of user experience and simulator sickness?
• RQ3 What are the trade-offs in fidelity, adaptability, and comfort between humanoid and human haptic feedback?
• RQ4 Is a smoothed, continuous synthesis better than a threshold-based approach for mapping g-forces to whole-body chair motion?

Key findings

• Humanoid feedback improved immersion and realism relative to no feedback and controller-only feedback.
• Humanoid feedback increased enjoyment but slightly reduced comfort due to higher simulator sickness in some users.
• Humanoid feedback yielded higher hedonic quality, while human feedback achieved higher pragmatic quality, highlighting a trade-off between consistency and adaptability.
• Latency analysis showed the pipeline stayed under 40 ms, with most time spent on inverse kinematics, indicating suitability for real-time haptic rendering.
• Interviews highlighted fidelity (realism), adaptability (predictable timing vs. human nuance), and versatility as core themes for humanoid-mediated haptics.
• Participants preferred humanoid or human feedback over controller or no feedback in overall rankings.