OrbitText: A Hybrid Prediction System for Efficient and Effortless Text Entry in Virtual Reality

Zihao Li, Dongdong Weng*, Xiaonuo Dongye, Jie Hao, Xiangyu Qi

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Text entry in virtual reality (VR) is a common task, however, it is often hindered by low typing efficiency and physical fatigue. To address these challenges, we present OrbitText, a novel text entry system for VR environments. OrbitText features a dual-handed, minimal-motion 3D interface and integrates an innovative typing prediction model that combines large language models (LLMs) with n-gram models. Our design aims to minimize physical strain while enhancing typing efficiency. User studies demonstrate that OrbitText significantly outperforms mainstream virtual keyboards, achieving higher efficiency, reduced fatigue, and a lower perceived task load.

Original languageEnglish
Title of host publicationProceedings - 2025 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops, VRW 2025
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1410-1411
Number of pages2
ISBN (Electronic)9798331514846
DOIs
Publication statusPublished - 2025
Event2025 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops, VRW 2025 - Saint-Malo, France
Duration: 8 Mar 202512 Mar 2025

Publication series

NameProceedings - 2025 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops, VRW 2025

Conference

Conference2025 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops, VRW 2025
Country/TerritoryFrance
CitySaint-Malo
Period8/03/2512/03/25

Keywords

  • Human computer interaction
  • Human-centered computing
  • Interaction techniques
  • Text input

Fingerprint

Dive into the research topics of 'OrbitText: A Hybrid Prediction System for Efficient and Effortless Text Entry in Virtual Reality'. Together they form a unique fingerprint.

Cite this