Visual Augmentation and Multimodal Assistive Technology for People With Low Vision: A Literature-Driven Design Study

Abstract

This thesis investigates how visual augmentation and multimodal assistive technology can be combined to support visual exploration for people with low vision. Existing assistive solutions either require native installation, lack real-time camera processing, or provide single capabilities in isolation; the dominant framing in the literature has been navigation rather than the more frequent everyday challenge of reading labels, identifying objects, and engaging with the visual environment in place.

Three interlocking methods were used: a narrative literature review, Research through Design across ten development phases, and semi-structured expert interviews with three practitioners in vision science and accessibility, analysed using Braun and Clarke's thematic framework.

The study produces three findings. A four-feature web-based prototype combining real-time contrast enhancement, AR text overlay, on-device OCR with audio output, and AI-assisted scene description is technically feasible on standard mobile browsers without installation. The deployment context creates a substantial gap between published benchmark performance and what runs reliably in a browser. The expert interviews generated a design-extending finding: scene description should support interactive targeted querying rather than exhaustive listing. The thesis contributes a working prototype, a documented design framework, and a conceptual reframing of visual augmentation as exploration rather than navigation. User testing with people with low vision is the essential next step.