The process of packing—organizing products within a container to optimize available space—is often performed manually without technological assistance, which can lead to inconsistent efficiency and occasional errors. While augmented reality (AR) has been explored to support packing tasks, prior research has largely relied on static final-pose overlays and has rarely integrated advanced packing optimization algorithms. This paper presents OPTAR, a novel packing assistance system that combines a 3D packing optimization algorithm with trajectory-based AR pose guidance, deployed on the Microsoft HoloLens 2. We describe the system's design, implementation, and evaluation through experiments comparing two methods: a packing operation with and without OPTAR. Results indicate that OPTAR assistance significantly improves space utilization and eliminates misplaced items, though with longer task completion times during initial use. To strengthen the generalizability of our findings to industrial applications, future work will extend testing to real industrial environments, involve professional packing operators, and larger sample sizes.
The communication barrier between deaf and hearing individuals remains a global challenge in academic, professional, and social contexts. Although numerous countries officially recognize their respective national sign languages, the availability of automatic, real-time, and linguistically accurate translation systems from spoken into sign languages remains limited in both coverage and quality. This paper reviews the key technologies, limitations, and interdisciplinary opportunities on computational approaches for real-time spoken-to-sign language translation. Following a systematic search of IEEE Xplore, Scopus, ACM Digital Library, and SciELO databases conducted in accordance with PRISMA guidelines, covering research published between 2020 and 2024, and applying predefined inclusion and exclusion criteria, we identified fifteen peer-reviewed studies. The analyzes of key enabling technologies, including automatic speech recognition (ASR), natural language processing (NLP), 3D avatar animation, and emerging applications in virtual reality (VR), identified major linguistic and technical challenges, such as the representation of non-manual signals, multimodal grammar processing, and the scarcity of high-quality datasets. The results revealed the following four dominant technological approaches: ASR-NLP pipelines, avatar-based rendering systems, VR/AR learning tools, and sensor-based gesture recognition platforms. Major limitations observed in these studies include low representation of non-manual signals, restricted vocabulary coverage, and high hardware dependency.
This paper presents an analysis of two immersive educational simulators developed at the [Anonymized]: SimCom and Mediation VR. SimCom simulates medical consultations using screen-based avatars controlled in real time by students or instructors, while Mediation VR is a virtual reality experience in which law students mediate conflicts between AI-driven avatars. Both tools adopt a role-play approach to foster communication and interpersonal skills. The study examines their pedagogical and technical characteristics and includes qualitative feedback from professors who apply them in the classroom. The findings indicate shared benefits, such as increased student participation and fluency in communication-based tasks, as well as implementation-specific challenges. Based on feedback, SimCom was restructured to improve usability and scalability. For Mediation VR, proposals are outlined to expand expressiveness, enable real-time feedback, and improve customization flexibility. The study reinforces the educational potential of immersive simulators and their adaptability to diverse academic contexts.
Virtual Reality (VR) is increasingly employed in healthcare training because it can create interactive three-dimensional learning environments that minimize risk to patients. As VR-based educational interventions become more widespread, evaluating their learning outcomes is becoming a critical focus for researchers and practitioners alike. Although there are a few other studies which verified VR learning in healthcare, they had as their subject the effectiveness of VR itself, not the methods which are used to evaluate learning. This systematic review examined 72 studies (selected from an initial pool of 22773 retrieved from IEEE Xplore, Scopus, Springer, ACM, and PubMed) that span a range of healthcare domains, with a primary focus on surgery and rehabilitation. The studies demonstrated a broad diversity of evaluation methods, including objective measures, subjective questionnaires, and AI-assisted assessments. While the overall methodological quality was generally high, the significant variability in evaluation approaches creates challenges for cross-study comparison and synthesis. The findings underline the necessity for validated, VR-specific assessment frameworks to facilitate the development and expansion of VR applications in lesser-explored areas of healthcare education.
Immersive technologies enhance training and performance evaluation in fields like medical education. By enabling detailed skill assessment through immersive analytics, Virtual reality (VR) systems also allow instructors to capture and study user interactions in simulated environments, linking theoretical metrics to real-world skill evaluations via behavioral visualization. This paper proposes a system for collecting, storing, and reconstructing user interaction data in medical VR simulations, recording hand movements, 6-DoF positioning, and object manipulation. Data is synchronized to a cloud platform, where a reconstruction process generates a real-time ’ghost avatar’ replay with spatial fidelity. Instructors and trainees review sessions immersively using interactive controls to analyze procedural steps, clinical decisions, and deviations from protocol, such as incorrect instrument handling, missteps in sequence, or safety violations within the original spatial context. To validate the tool, a user experiment evaluates the impact of 3D reconstructed scenes on perceived learning and self-assessment. Participants perform predefined tasks, e.g., instrument interaction, and qualitatively compare traditional debriefing with immersive replay, reporting on comprehension and perceived improvements. Results demonstrate that spatially grounded replay enhances perceived feedback precision and can improve training effectiveness compared to conventional methods. Cloud storage ensures scalability for asynchronous review, broadening adoption in resource-constrained settings.
