How Accurate is the Positioning in VR? Using Motion Capture and Robotics to Compare Positioning Capabilities of Popular VR Headsets

TL;DR

This work tackles the problem of measuring VR headset positioning accuracy by recording real head trajectories with an optical motion capture system and then reproducing them with a UR5e robot. The authors develop a reproducible pipeline that includes Unity–OptiTrack calibration, TCP–Unity hand‑eye calibration, and a MoveIt/Gazebo‑based simulation to plan and execute robot trajectories that mirror human head motion. They evaluate Meta Quest 2 and Quest Pro across calibration, real‑world gameplay, and long‑duration trajectories using Absolute Pose Error metrics, finding both headsets to exhibit high accuracy with only modest differences in most tests. The approach yields a practical, vendor‑neutral framework for VR tracking evaluation with potential applications in manufacturing and professional training, and the data/software are made available to the community.

Abstract

In this paper, we introduce a new methodology for assessing the positioning accuracy of virtual reality (VR) headsets, utilizing a cooperative industrial robot to simulate user head trajectories in a reproducible manner. We conduct a comprehensive evaluation of two popular VR headsets, i.e., Meta Quest 2 and Meta Quest Pro. Using head movement trajectories captured from realistic VR game scenarios with motion capture, we compared the performance of these headsets in terms of precision and reliability. Our analysis revealed that both devices exhibit high positioning accuracy, with no significant differences between them. These findings may provide insights for developers and researchers seeking to optimize their VR experiences in particular contexts such as manufacturing.

Figures (9)

• Figure 1: Real-world head movement data acquisition with motion capture setup (OptiTrack) during VR gameplay.
• Figure 2: Schematic view of the structure of the proposed positioning accuracy assessment system.
• Figure 3: Meta Quest Pro with OptiTrack markers attached.
• Figure 4: Artificial head with the Meta Quest Pro attached, mounted on the Universal Robots UR5e manipulator tip (up), and the real-world test environment as seen from the perspective of the mounted headset (down).
• Figure 5: Diagram illustrating the relation between systems when determining TCP poses of interest $T_i$ for each OptiTrack pose of the real-world head trajectory $O_i$. The diagram consists of two fixed elements: $\mathbf{A}={^{\rm TCP}}\mathbf{T}_{\rm Unity} \cdot {^{\rm Unity}}\mathbf{T}_{\rm Opti}$, representing the transformation between the TCP and OptiTrack's Rigid Body corresponding to the VR headset, and $\mathbf{B}$, which denotes the initial pose of the TCP in its coordinate system. $\Delta{O}_{i}$ and $\Delta{H}_{i}$ represent the changes in poses within their respective systems, relative to the starting point, for the trajectory recorded by OptiTrack and the target TCP trajectory. The final TCP poses of interest can be determined with the formula $T_i = \mathbf{B} \cdot \Delta{H}_{i}$, where $\Delta{H}_i=\mathbf{A} \cdot \Delta{O}_i \cdot \mathbf{A}^{-1}$.