Effective collaboration in Virtual Reality depends on users' ability to accurately perceive the affordances of others within a shared space. This paper presents the first empirical investigation of how people judge others' reaching capabilities in Immersive Virtual Environments (IVEs). We examined whether observers can reliably estimate the maximum horizontal reach distance of virtual actors, how different actor representations influence these perceptual judgments, and whether calibration feedback can improve estimation accuracy. Participants (N=35) judged the reachability of objects for virtual actors across three visual fidelity conditions: high-fidelity, low-fidelity, and end-effector-only representations. Results demonstrate that users can successfully estimate others' horizontal reachability in IVEs, though with overestimation. Calibration through action feedback significantly improved estimation accuracy, with overestimation decreasing across trials. Visual fidelity showed no main effect on initial reachability judgments but significantly affected postcalibration performance, with low-fidelity representations demonstrating significantly better calibration outcomes compared to high-fidelity avatars, while end-effector-only representations showed no significant difference from either condition. The findings contribute to the social perception literature in Virtual Reality (VR) and to provide design guidelines for avatar systems in collaborative virtual environments, where spatial coordination and shared action planning are critical.

The poor transparency of occlusion-capable optical see-through head-mounted displays (OC-OSTHMDs) deteriorates the visibility of the real scene, hindering the practical application of the devices. Previous works mitigate the issue by upgrading the transmittance of the spatial light modulator (SLM). However, the strategy soon reaches a limit because further optimization requires improving the transmittance of all optical elements, e.g., lenses and beam splitters. Moreover, pixelated occlusion usually relies on polarizing the real scene light, inevitably cutting the input optical power by half. To overcome this limitation, we propose a mask balancing method that improves real-scene brightness through polarization blending. Specifically, the s-polarized component, which passes through the optical system to provide occlusion-capable vision, is blended with the p-polarized component, which bypasses the system to preserve the raw view of the real scene. The blending is realized by simply modulating the cross-angle between a polarizing beam splitter and a linear polarizer, benefiting the robustness and versatility of the proposed method. We introduce a perception-driven blending approach, where the cross-angle is optimized in real-time to balance the visibility of the real scene and the texture and lighting of the virtual object. A benchtop prototype is built. A user study with 12 participants is conducted to quantify the visibility threshold of the texture and lighting of virtual objects. Then, a user study with 12 participants proves that the proposed method improves the visibility of the real scene while keeping a good appearance of the virtual object. We believe the proposed method is an important step toward developing practical solutions for OC-OSTHMDs.

Recent advances in leveraging large-scale Internet photo collections for 3D reconstruction have enabled immersive virtual exploration of landmarks and historic sites worldwide. However, existing methods primarily focus on visual appearance reconstruction, often overlooking the interactive semantic understanding of these 3D scenes (e.g., identifying specific building parts or scene details), which remains largely confined to browsing static text-image pairs. Therefore, can we draw inspiration from 3D in-the-wild reconstruction techniques and use unconstrained photo collections to create an immersive approach for comprehensive 3D scene understanding beyond mere visual appearance? To this end, we extend language embedded 3D Gaussian splatting (3DGS) and propose a novel framework for open-vocabulary scene understanding from unconstrained photo collections. Specifically, we first render multiple appearance images from the same viewpoint as the unconstrained image with the reconstructed radiance field, then extract multi-appearance CLIP features and two types of language feature uncertainty maps-transient and appearance uncertainty-derived from the multi-appearance features to guide the subsequent optimization process. Next, we propose a transient uncertainty-aware autoencoder, a multi-appearance language field 3DGS representation, and a post-ensemble strategy to effectively compress, learn, and fuse language features from multiple appearances. Finally, to quantitatively evaluate our method, we introduce PT-OVS, a new benchmark dataset for assessing open-vocabulary segmentation performance on unconstrained photo collections. Experimental results show that our method outperforms existing methods, delivering accurate open-vocabulary segmentation and enabling applications such as interactive roaming with open-vocabulary queries, architectural style pattern recognition, and 3D scene editing. Visit our project page at Project Page.

The aviation industry faces increasing pilot training demands, and reliance on conventional Full Flight Simulators (FFS) limits training capacity. Virtual Reality (VR) offers scalable, remote training opportunities, but its role as a complement to FFS requires empirical validation. In collaboration with Singapore Airlines (SIA) instructor pilots, we developed a VR training prototype for a visual approach into Gimhae International Airport, emphasizing Competency-Based Training and Assessment (CBTA)-based Threat and Error Management (TEM). An empirical study with 39 SIA Boeing 737-MAX 8 type-rated first officers evaluated VR against FFS. An equivalence analysis showed that VR achieved performance outcomes comparable to FFS in 13 of the 16 Observable Behaviors (OBs) and across 4 Competencies. In addition, a comparative analysis indicated measurable performance improvements when VR was used to supplement FFS training. Our results suggest VR can meaningfully complement FFS in targeted competency areas, with future work required to assess its broader integration across additional scenarios and performance metrics.

Postural Instability Theory (PIT) proposes that individuals who are naturally unstable on their feet are more susceptible to cybersickness. We hypothesize that this relationship extends to locomotive VR, such that people who exhibit greater instability when walking without VR will also be more susceptible to cybersickness in a locomotive VR setup. To test this, we analyzed participants' natural walking kinematics alongside their cybersickness responses and kinematic patterns during mobile VR use. Our results showed that vertical Center of Mass movement during pre-VR walking showed promise for identifying individuals susceptible to cybersickness. Spatial stability metrics emerged as stronger predictors of cybersickness than time-series measures, suggesting that spatial characteristics of gait may be more informative indicators of susceptibility in mobile VR contexts. These findings highlight the importance of accounting for baseline postural stability when designing and personalizing mobile VR experiences.

Mixed Reality (MR) environments are reshaping education by fostering engagement and motivation. Within such environments, Intelligent Virtual Agents (IVAs) can serve as personalized guides, tutors, or collaborative partners. In this research, we present a novel IVA for education that perceives the learner's environment, adapts its guidance to their progress, and interacts with both the learner and the real world. Our multimodal IVA integrates speech capabilities powered by large language models (LLMs) with computer vision for dynamic scene understanding, allowing it to converse naturally with learners and provide adaptive support based on their progress. To evaluate our IVA, we implemented a Scratch programming scenario using the Magic Leap 2 headset. We conducted a user study with 24 university students (12 male, and 12 female) who completed the same tasks using three systems: basic AR cues, a chatbot with AR cues, and our IVA. Quantitative data were obtained using standardized questionnaires such as the Technology Acceptance Model (TAM), the User Engagement Scale (UES), and the System Usability Scale (SUS), alongside qualitative feedback. The results revealed that although the IVA system took a longer time compared to other systems, it fostered greater motivation and engagement, achieved the highest SUS score, and showed the strongest TAM correlations. Participants perceived IVA as the most valuable system, highlighting the potential of multimodal embodied agents as virtual tutors in MR.

Immersive virtual reality (VR) applications demand accurate, temporally coherent full-body pose tracking. Recent head-mounted camera-based approaches show promise in egocentric pose estimation, but encounter challenges when applied to VR head-mounted displays (HMDs), including temporal instability, inaccurate lower-body estimation, and the lack of real-time inference. To address these limitations, we present EgoPoseVR, an end-to-end framework for accurate egocentric full-body pose estimation in VR that integrates headset motion cues with egocentric RGB-D observations through a dual-modality fusion pipeline. A spatiotemporal encoder extracts frame- and joint-level representations, which are fused via cross-attention to fully exploit complementary motion cues across modalities. A kinematic optimization module then imposes constraints from HMD signals, enhancing the accuracy and stability of pose estimation. To facilitate training and evaluation, we introduce a large-scale synthetic dataset of over 1.8 million temporally aligned HMD and RGB-D frames across diverse VR scenarios. Experimental results show that EgoPoseVR outperforms state-of-the-art egocentric pose estimation models. A user study in real-world scenes further shows that EgoPoseVR achieved significantly higher subjective ratings in accuracy, stability, embodiment, and intention for future use compared to baseline methods. These results show that EgoPoseVR enables robust full-body pose tracking, offering a practical solution for accurate VR embodiment without requiring additional body-worn sensors or room-scale tracking systems.

Visual perception plays a critical role in detecting changes within immersive Virtual Reality (VR) environments. However, as visual complexity increases, perceptual performance declines, making it more difficult to detect changes quickly and accurately. This study examines how visual features, known for facilitating preattentive processing, impact a change detection task in immersive 3D environments, with a focus on visual complexity, object attributes, and spatial proximity. Our results demonstrate that preattentive processing enhances change detection, particularly when the altered object is spatially isolated and not perceptually grouped with similar surrounding objects. Changes to isolated objects were detected more reliably, suggesting that perceptual isolation reduces cognitive load and draws more attention. Conversely, when a changed object was surrounded by visually similar elements, participants were less likely to detect the change, indicating that perceptual grouping hinders individual object recognition in complex scenes. These results provide guidelines for designing VR applications that strategically utilize spatial isolation and visual features to improve the user experience.

Mitigating motion sickness with non-invasive cortical vestibular brain stimulation (vBS) is promising because it does not require redesigning VR content or altering the motion profile of a moving platform. Instead, it aims to enhance individuals' intrinsic resistance to motion sickness. However, the mechanism underlying its online efficacy remains unclear (Here, "online efficacy" means the observed motion sickness reduction during vBS not before or after vBS). Motivated by lower motion sickness susceptibility observed in congenital nystagmus patients, we hypothesized that reductions in motion sickness under vBS condition are associated with increased eye movement instability. Leveraging sham-controlled, sex-balanced eye-tracking data from an open-access database, this study presents the first evaluation of a 20 Hz-vBS approach targeting the vestibular cortices as a countermeasure to motion sickness, using spatiotemporal eye movement metrics derived from eye-tracking data collected in VR. Results show that reduced motion sickness under active vBS indeed increased eye movement instability temporarily and spatially, compared to sham vBS. We also found that the temporal metrics was more strongly associated with female participants, whereas the spatial metric was more strongly associated with male participants.

The Simulator Sickness Questionnaire (SSQ) has become a standard tool for quantifying the severity and distribution of discomfort symptoms in virtual reality (VR) research. Despite its straightforward administration, the use of the SSQ also comes with significant challenges, including response subjectivity, strict threshold values based on a military reference population, and confounding factors influencing the results. To demonstrate the adverse interplay of these issues, we asked three cohorts of students to fill in a SSQ after having attended a 90-minute lecture of our teaching program. Although students were not exposed to any form of VR experience, the resulting SSQ scores were indistinguishable from VR studies and extended far beyond the originally defined threshold of a "bad simulator", with 88.1% of TS scores being larger and 25.4% even exceeding thrice this value. We compare our results to alternative scoring systems of the SSQ proposed in the literature and suggest implications for future experimental designs involving the quantification of sickness symptoms. In summary, our results motivate to exert caution when interpreting the results of the SSQ in the context of a VR study; participants might just have attended a lecture prior to the experiment.