Publications

Evaluates multilingual AR+BIM interfaces for construction training. Spanish-speaking participants completed a multi-step electrical installation task using AR guidance in Spanish or English, enabling comparison of task performance, workload, usability, and language-related factors.

Reviews benchmarks and methods for continuous hand gesture recognition, proposes a taxonomy for organizing evaluation, and identifies limitations in dataset coverage, evaluation protocols, real-time recognition, and reproducibility.

Studies response-delay mitigation in free-form spoken interaction with LLM-powered virtual agents and evaluates design strategies intended to improve conversational responsiveness and the resulting user experience.

Introduces ALM-Bench, a benchmark for evaluating large multimodal models across 100 culturally diverse and low-resource languages, supporting systematic analysis of multilingual and multicultural model understanding.

Surveys object-selection techniques used in consumer VR applications, organizes them through a taxonomy, and compares deployed practice with methods emphasized in academic research to identify gaps between research and real-world use.

Examines the relationship between users' self-reported confidence and their actual accuracy when estimating distances in virtual reality, using blindwalking tasks to compare subjective judgments with objective performance.

Compares manual and automatic point-of-release mechanics for VR throwing across multiple input configurations and object types in a controlled user study, providing design guidance based on both performance and participant preferences.

Presents a preliminary evaluation of five input configurations and multiple throwable objects for VR throwing, comparing performance and participant preferences and motivating the later full study of point-of-release mechanics.

Develops a taxonomy of collaborative human-centered XR tasks from three decades of literature, organizing task actions and properties to clarify design choices and support comparison across collaborative immersive systems.

Develops a GDPR-grounded item bank for measuring privacy preferences by translating regulatory protections into user-facing statements and organizing the resulting items into a structured set of privacy themes for future measurement and validation.

Presents a locally deployed LLM-powered conversational-avatar system integrating automatic speech recognition, speech synthesis, lip synchronization, and response-status indicators, together with reference architectures for state-machine and retrieval-grounded conversational behavior.