Investigating A Geometrical Solution to the Vergence-Accommodation Conflict for Targeted Movements in Virtual Reality

TL;DR

This study probes the vergence-accommodation conflict (VAC) in VR through a vision-based geometrical model that predicts a constant vergence offset distorting binocular geometry and causing movement undershooting. It validates the model in Experiment 1, showing VAC chiefly disrupts online disparity matching rather than feedforward distance planning, and introduces a shader-based transformation that compensates for VAC, achieving about a 30% improvement in movement accuracy in Experiment 2. The work demonstrates a practical, hardware-free method to mitigate VAC in HMD-based tasks, though effectiveness varies across individuals and distances, underscoring the need for adaptive personalization in future VR systems. Overall, the approach leverages intrinsic VAC geometry to reduce perceptual-motor errors, with potential impact for precision VR applications such as surgical training and remote robotics.

Abstract

While virtual reality (VR) holds significant potential to revolutionize digital user interaction, how visual information is presented through VR head-mounted displays (HMDs) differs from naturalistic viewing and interactions in physical environments, leading to performance decrements. One critical challenge in VR development is the vergence-accommodation conflict (VAC), which arises due to the intrinsic constraints of approximating the natural viewing geometry through digital displays. Although various hardware and software solutions have been proposed to address VAC, no commercially viable option has been universally adopted by manufacturers. This paper presents and evaluates a software solution grounded in a vision-based geometrical model of VAC that mediates VAC's impact on movement in VR. This model predicts the impact of VAC as a constant offset to the vergence angle, distorting the binocular viewing geometry that results in movement undershooting. In Experiment 1, a 3D pointing task validated the model's predictions and demonstrated that VAC primarily affects online movements involving real-time visual feedback. Experiment 2 implemented a shader program to rectify the effect of VAC, improving movement accuracy by approximately 30%. Overall, this work presented a practical approach to reducing the impact of VAC on HMD-based manual interactions, enhancing the user experience in virtual environments.

Figures (7)

• Figure 1: Visualizations of the binocular viewing geometry. (a) The geometry under a normal viewing condition. The binocular disparity associated with the target (black circle) can be derived as the difference of the visual angle of the target relative to the fixation (black cross), $\delta = \alpha_L - \alpha_R$, which is equivalent to the difference between the vergence angle $\phi$ and the visual angle subtended by the target $\tau$, $\delta = \phi - \tau$. Importantly, under a normal viewing condition, vergence (black, solid lines) and accommodation (green areas) are congruent and correspond to the same depth. (b) The perturbed binocular viewing geometry with the mediation of a screen. The screen draws the accommodation to a fixed location, closer than the fixation. As a result, the vergence is also drawn inward, resulting in a larger vergence angle, $\hat{\phi}$. (c) The viewing geometry with the transformation that renders the target further than it is to offset the depth compression induced by the distorted vergence angle.
• Figure 2: An illustration of the procedure for Experiment 1 from the participant’s point of view. The trapezoid represents the tabletop on which stimuli were displayed extending into the distance. The hand represents the virtual hand with which the participants performed the 3D pointing movement in VR. The participants could either see the target (Online Guidance) or not (Feedforward) when performing the movement after the beep.
• Figure 3: The mean distance errors from the behavioral experiment (solid lines) and model predictions (gray, dashed lines) as a function of target distance for different feedback conditions in Experiment 1. Model predictions were based on the optimal versions of the model for each feedback condition. Error bars represent 95% confidence intervals.
• Figure 4: Predicted endpoint errors for the original and transformed virtual environment based on Equation \ref{['eq: shader_z']} as a function of target lengths.
• Figure 5: The participant's view through the HMD during Experiment 2 without (left) and with (right) the transformation applied to the virtual environment.