Logo do repositório
 
A carregar...
Miniatura
Publicação

HexGaze: Hexagonal Virtual Keyboard with Word Prediction for Eye Typing

Utilize este identificador para referenciar este registo.
Nome:Descrição:Tamanho:Formato: 
TM_Robert_Cachapa.pdf13.43 MBAdobe PDF Ver/Abrir

Resumo(s)

Although eye typing has existed for several years, many gaze-based systems still struggle when using low-cost eye trackers, resulting in slow typing speeds, high error rates, and increased user fatigue and discomfort. These challenges highlight the need for continued improvement and innovation in eye typing solutions. This work introduces and evaluates HexGaze, a gaze-based text entry technique designed to be comfortable, fast, and tolerant of low-cost, imprecise eye trackers. The system features a unique virtual keyboard with a hexagonal layout that keeps user interaction centered on the screen, where eye trackers accuracy is the highest. It combines fluid, gesture-based selection with a dwell-time method to allow users the freedom to chose the most comfortable approach. It also integrates a word prediction system for both current-word completion and nextword suggestions to enhance typing efficiency. Two user studies were conducted to evaluate HexGaze. The first study compared it against two existing techniques, WordPop and Dasher. The results showed that HexGaze achieved higher typing speeds (5.96 WPM), lower error rates, and greater keystroke efficiency than WordPop. Although HexGaze also outperformed Dasher, the difference was not statistically significant, suggesting a trend in favour of HexGaze but not a conclusive advantage. Participants also rated it as the most comfortable, least fatiguing, and easiest to learn. The second study analysed the learning curve over eight sessions, revealing a strong and consistent improvement in performance. Typing speed nearly doubled, from an average of 5.59 WPM (first session) to 9.94 WPM (last session). Efficiency (KSPC) also increased substantially from the first session to the last. Crucially, error rates remained stable and eye fatigue remained low throughout, demonstrating the technique’s suitability for prolonged use. Finally, the findings demonstrate that HexGaze offers a superior balance of speed, accuracy, and user comfort. Its intuitive design and strong learning curve make it a promising, user friendly, and practical solution for eye typing.

Descrição

Tese de mestrado, Engenharia Informática, 2026, Universidade de Lisboa, Faculdade de Ciências

Palavras-chave

Eye Typing HexGaze Text entry Gestures Dwell Time

Contexto Educativo

Citação

Projetos de investigação

Unidades organizacionais

Fascículo