A Tool for Estimating Success Rates of Raycasting-Based Object Selection in Virtual Reality

TL;DR

This paper addresses the challenge of quantifying VR UI usability by estimating raycasting-based object selection success rates. It builds a VR-specific endpoint-distribution model using a bivariate Gaussian with mean $\boldsymbol{\mu}$ and covariance $\boldsymbol{\Sigma}$, enabling SR to be computed as the probability mass within a target region. The authors implement an open-source Unity editor extension that applies this model to real-time UI elements, and validate the approach with a data collection study ($n=18$, $8640$ trials) and two developer user studies, achieving an average SR estimation error around $2$–$4\%$ and strong predictive power ($R^2\approx0.99$). The work demonstrates practical, quantitative guidance for VR UI design, potentially reducing design iterations by providing objective targets for element size and placement and bridging research insights with development practice.

Abstract

As XR devices become widespread, 3D interaction has become commonplace, and UI developers are increasingly required to consider usability to deliver better user experiences. The HCI community has long studied target-pointing performance, and research on 3D environments has progressed substantially. However, for practitioners to directly leverage research findings in UI improvements, practical tools are needed. To bridge this gap between research and development in VR systems, we propose a system that estimates object selection success rates within a development tool (Unity). In this paper, we validate the underlying theory, describe the tool's functions, and report feedback from VR developers who tried the tool to assess its usefulness.

Figures (4)

• Figure 1: (A) Angular representations of target width $W$ and movement amplitude $A$. (B) Participants sequentially selected targets following the arrowed path.
• Figure 2: Observed success rates for spherical targets and success rates estimated using the endpoint distribution derived from our experiment.
• Figure 3: Left: Inspector window of the proposed success-rate estimation tool implemented as a Unity editor extension. Developers specify analysis settings such as the reference camera, target shape (e.g., circular), coordinate system (e.g., world space), and optional model parameters. Right: Unity scene view showing a VR UI canvas. When a UI element is selected, the tool overlays its apparent angular size (target width $W$ in degrees, e.g., $0.95^\circ$) and the predicted selection success rate (e.g., 70.41%) near the object.
• Figure 4: (A) UI before adjustment. (B) UI after adjustment by a participant. The canvas was attached to the left-hand controller.