GazeTheWeb: A Gaze-Controlled Web Browser
Raphael Menges, Chandan Kumar, Daniel Müller, Korok Sengupta · 2017 · Proceedings of the 14th International Web for All Conference (W4A) · doi:10.1145/3058555.3058582
Summary
This demonstration paper presents GazeTheWeb, an open-source web browser designed to be operated entirely through eye gaze input, targeting people with severe motor impairments who cannot use conventional mouse and keyboard input. Built on the Chromium Embedded Framework (CEF), the browser takes a fundamentally different approach from existing gaze-based solutions: rather than emulating mouse and keyboard movements with eye tracking (as tools like OptiKey do), GazeTheWeb redesigns the entire browsing interface around gaze interaction paradigms. The system works by extracting interactive web elements from pages — links, text input fields, scrollable sections, select fields — using JavaScript-induced DOM callbacks, then representing them with explicit or implicit indicators optimized for gaze-based activation. Interface components are designed with size, shape, appearance, and feedback characteristics specifically calibrated to compensate for the inherent accuracy limitations of eye tracking hardware. The browser supports all core browsing operations including scrolling (via gaze-activated sensors at page top and bottom that accelerate with fixation duration), hyperlink navigation (using continuous zooming to resolve imprecise gaze coordinates into precise click targets), text input (via a virtual keyboard with dwell-time activation), and full tab management including bookmarks and URL editing.
Key findings
A comparative study with 11 participants (4 female, 7 male) showed GazeTheWeb significantly outperformed the conventional approach of using OptiKey mouse/keyboard emulation with Google Chrome. Average task completion time was 252.18 seconds with GazeTheWeb versus 424.15 seconds with OptiKey (p=0.0023) — a 40% speed improvement. System Usability Scale (SUS) scores were 83.86 for GazeTheWeb versus 57.96 for OptiKey, placing GazeTheWeb in the "excellent" usability range while OptiKey scored below average. NASA-TLX workload scores were also lower (better) for GazeTheWeb at 43.0 versus 54.5 for OptiKey, indicating reduced cognitive and physical effort. The key technical innovation is the continuous zooming approach for link selection: since eye tracker accuracy is insufficient for directly selecting small hyperlinks, the browser progressively zooms into the area the user is looking at while highlighting extracted links, performing a click only when the zoom level allows precise coordinate identification. Clinical studies were underway with users who have neuromuscular disease, spinal cord injury, and Parkinson's disease.
Relevance
GazeTheWeb demonstrates that purpose-built assistive interfaces can dramatically outperform generic input emulation approaches. This has broad implications for assistive technology design: rather than retrofitting accessibility onto existing interfaces through input translation layers, building natively accessible interfaces from the ground up yields substantially better usability. For web developers, the research highlights that well-structured HTML with properly identified interactive elements (links, form fields, buttons) is essential for enabling alternative input methods — the browser relies on being able to extract and classify these elements from the DOM. The open-source nature of the project (hosted on GitHub under the EU-funded MAMEM project) makes it a practical resource for people with severe motor impairments. The work is particularly relevant for conditions like ALS, muscular dystrophy, and high-level spinal cord injuries where eye gaze may be the only reliable voluntary movement available.
Tags: eye tracking · gaze interaction · web browser · motor impairment · assistive technology · alternative input