HumanoidTurk: Expanding VR Haptics with Humanoids for Driving Simulations

TL;DR

HumanoidTurk investigates using a general-purpose humanoid robot as a whole-body haptic medium for VR driving. The approach translates in-game g-forces into synchronized chair motion via a two-path synthesis pipeline and evaluates real-time performance and user experience through pilot and main studies. Results show humanoid feedback enhances immersion, realism, and enjoyment relative to no feedback or controller vibrations, but can increase simulator sickness and affect comfort; human-delivered feedback remains subtly more adaptable. The work identifies fidelity, adaptability, and versatility as core themes and argues for humanoids as a promising, versatile modality for future VR haptics, with opportunities spanning training, rehabilitation, and entertainment.

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.

Figures (6)

• Figure 1: Pipeline of our smoothed synthesis approach. In this example, when the participant drives over a curb while cornering, the LPF of lateral g-force generates a sway feedback. In contrast, the combination of LPF and HPF of longitudinal g-force triggers vibration feedback.
• Figure 2: Experimental setup for the four feedback conditions. (a) No-Feedback, where the participant drives in VR using only visual and auditory cues, and (b–d) conditions in which vibration is delivered through the game controller: (b) Controller; (c) Humanoid+Controller, where the humanoid robot apply motion feedback to the chair; and (d) Human+Controller, where a human operator behind the participant apply motion feedback to the chair.
• Figure 3: The boxplot of the results for Pragmatic Quality and Hedonic Quality from UEQ-S. The Human+Controller condition achieved the highest PQ, while the Humanoid+Controller scored significantly higher in HQ, indicating superior enjoyment. The No-Feedback condition consistently scored lowest across both dimensions. (*p<.05, **p<.01, ***p<.001)
• Figure 4: The boxplot of the results for individual questionnaires. The Humanoid+Controller and Human+Controller conditions consistently achieved higher ratings for Immersion, Realism, Enjoyment, and Suitability compared to the baselines. However, a trade-off was observed in Comfort, where the Controller condition was rated higher than the Humanoid+Controller. (*p<.05, **p<.01, ***p<.001)
• Figure 5: Overall Preference. Participants showed a strong preference for conditions with motion feedback, with the Humanoid+Controller and Human+Controller conditions securing the majority of 1st and 2nd place rankings. In contrast, the No-Feedback condition was ranked last.