What do Blind and Low-Vision People Really Want from Assistive Smart Devices? Comparison of the Literature with a Focus Study
Bhanuka Gamage, Thanh-Toan Do, Nicholas Seow Chiang Price, Arthur Lowery, Kim Marriott · 2023 · Proceedings of the 25th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS '23) · doi:10.1145/3597638.3608955
Summary
This paper investigates whether the tasks and devices explored by researchers developing AI-based assistive smart devices actually align with what blind and low-vision (BLV) people want. The authors conducted a three-part study combining a scoping literature review with semi-structured interviews of 24 BLV participants. In the first study, they identified and analyzed 646 research papers published between 2020 and mid-2022 that investigated computer vision-based assistive technologies for BLV users. They categorized each paper by task (e.g., object detection, navigation, text recognition), device type (smartphones, smart glasses, body-mounted devices), interaction model (scene description, image enhancement, question answering), and whether BLV participants were actually involved in the research. The literature review revealed that the majority of studies focused on assistive products for handling objects and devices (42.7%), personal mobility (40.4%), and communication and information (32.2%). Smartphones and body-mounted devices dominated the research landscape. Critically, 82% of studies did not involve BLV participants at all, and only 38 of the 646 papers involved BLV people in the design or requirements gathering stage. In the second study, 24 BLV participants (14 blind, 10 low-vision) were interviewed about their top five desired tasks, preferred devices, and interaction modalities, and were asked to rank the usefulness of tasks identified in the literature.
Key findings
The correlation between researcher focus and BLV participant priorities was only weakly positive (r=.14, p=.24), revealing a significant misalignment. Text recognition was the most preferred task among participants (selected by 15 of 20), yet it was not proportionally represented in the literature. Obstacle detection had the most first-preference selections. Several tasks highly valued by participants—including filling paper forms, aerial obstacle detection, shop recognition, and empty seat detection—received little researcher attention. Conversely, heavily researched tasks like terrain detection, currency detection, and pedestrian detection were rated as relatively unimportant by participants. All 24 participants preferred a conversational agent (question-answering) interaction model, with 23 preferring voice input and all preferring speech output. Participants favored head-mounted devices for their hands-free operation and natural gaze alignment, though device preference varied by task and context. Studies that involved BLV participants during the design stage showed a significantly higher correlation (r=.27, p=.02) between researcher focus and user priorities, compared to studies with no BLV involvement (r=.13, p=.28), suggesting that participatory design leads to more relevant research.
Relevance
This study delivers a compelling evidence base for the importance of participatory design in assistive technology research. The finding that 82% of assistive AI studies exclude BLV participants entirely—and that this exclusion correlates with misaligned research priorities—should prompt researchers and funders to mandate user involvement from the earliest design stages. For practitioners and organizations developing assistive products, the ranked task lists and device preferences provide actionable guidance on what BLV users actually need versus what researchers assume they need. The strong preference for conversational interfaces and head-mounted devices points toward a future of AI-powered smart glasses functioning as personal assistants. The study also highlights under-explored but highly desired applications like form filling, empty seat detection, and shop navigation that represent opportunities for impactful development work.
Tags: blind and low vision · assistive technology · smart devices · computer vision · wearable technology · user needs · scoping review
Standards referenced: ISO 9999 · WHO ISO Assistive Technology Standards and Guidelines