Supporting Orientation of People with Visual Impairment

Abstract

In the field of assistive technology, large scale user studies are hindered by the fact that potential participants are geographically sparse and longitudinal studies are often time consuming. In this contribution, we rely on remote usage data to perform large scale and long duration behavior analysis on users of iMove, a mobile app that supports the orientation of people with visual impairments. Exploratory analysis highlights popular functions, common configuration settings, and usage patterns among iMove users. The study shows stark differences between users accessing the app through VoiceOver and other users, who tend to use the app more scarcely and sporadically. Analysis through clustering of VoiceOver iMove user interactions discovers four distinct user groups: 1) users interested in surrounding points of interest, 2) users keeping the app active for long sessions while in movement, 3) users interacting in short bursts to inquire about current location, and 4) users querying in bursts about surrounding points of interest and addresses. Our analysis provides insights into iMove's user base and can inform decisions for tailoring the app to diverse user groups, developing future improvements of the software, or guiding the design process of similar assistive tools. CCS Concepts •Human-centered computing → Accessibility design and evaluation methods; •Computing methodologies → Cluster analysis; •Social and professional topics → People with disabilities;

Insights on Assistive Orientation and Mobility of People with Visual Impairment Based on Large-Scale Longitudinal Data

“…Moreover, we believe that these results also provide insights for future mobility assistance applications. For example, the identification of a major cluster in our initial analysis [27], where users sporadically interact with i Move just to confirm their location, was one of the inspirations behind the virtual navigation in [22]. We suspected that those users already possessed a mental representation of their environment and used i Move for confirmation purposes.…”

“…We extend this approach by incorporating the notion of a session based on an inactivity threshold (as in Meier et al [34]). While alternative techniques that are tolerant to specific forms of permutable, redundant and omittable user-interaction patterns within sequences have been recently proposed [16], for comparison purposes we follow a similar approach to our initial analysis [27].…”

Mind Your Crossings

“…For example, iMove 3 is a commercial application that informs the user about the current address and nearby points of interest. Large scale analysis of iMove usage data from 4055 users of the system shows how users can be clustered in different user groups based on their interest in different types of spatial information [Kacorri et al 2016]. Other research prototypes, Crosswatch [Coughlan and Shen 2013] and ZebraLocalizer [Ahmetovic et al 2011], allow pedestrians with visual impairments to detect crosswalks with a smart-phone camera.…”

Hacking Blind Navigation

“…Potential avenues for future research may arise from the increasing efort to release large-scale usage datasets [8,14]. The availability of such datasets can promote new data-driven approaches to study and model user behavior, improve interface design, and support adaptive interfaces.…”