Design and Real-World Evaluation of Eyes-Free Yoga: An Exergame for Blind and Low-Vision Exercise
Kyle Rector, Roger Vilardaga, Leo Lansky, Kellie Lu, Cynthia L. Bennett, Richard E. Ladner, Julie A. Kientz · 2017 · ACM Transactions on Accessible Computing (TACCESS) · doi:10.1145/3022729
Summary
This paper presents Eyes-Free Yoga, an accessible exergame using the Microsoft Kinect that acts as a yoga instructor for people who are blind or have low vision, providing personalized auditory feedback based on real-time skeletal tracking. People with visual impairments face significant barriers to exercise participation due to inaccessible fitness environments, visual instructional materials, and lack of encouragement — contributing to higher rates of obesity and sedentary behaviour. Eyes-Free Yoga addresses these barriers by enabling independent, in-home yoga practice with audio-only guidance. The system was developed through an iterative design process involving yoga instructors, O&M specialists, and users with visual impairments. It supports six standing yoga poses across four complete workouts with progressive difficulty. The audio feedback system uses body-relative metaphorical instructions (e.g., "stretch your arms out to your sides, like a tightrope walker's pole") rather than abstract directional commands, making pose guidance understandable without visual demonstration. Two studies were conducted: a controlled laboratory study with 16 blind and low-vision participants evaluating personalized versus generic feedback, and an 8-week in-home deployment with 4 participants using an ABAB single-case experimental design comparing a Standard version against an Enhanced version with motivational techniques including badges, levels, progress tracking, and musical achievement cues.
Key findings
In the laboratory study, participants strongly preferred personalized feedback that told them specifically how to adjust their body position over generic encouragement, confirming that real-time corrective guidance based on skeletal tracking is both feasible and valued. The 8-week deployment study demonstrated sustained engagement: participants practiced an average of 17 hours across 24 days, approaching the American Heart Association's recommended 150 minutes per week of exercise. The Enhanced version with motivational techniques improved user experience scores and, at a descriptive level, increased exercise frequency and duration compared to the Standard version, though with only 4 participants statistical significance was limited. Three of four participants showed statistically significant increases in at least one measure during Enhanced phases. The ABAB design proved valuable for accessibility research where participant pools are small — single-case experimental designs can achieve statistical power even with very few participants. Key design insights included the effectiveness of metaphorical instructions for eyes-free guidance, the importance of verbally communicating game status that sighted users would see visually (progress bars, badges, levels), and the value of using mainstream consumer hardware (Kinect at ~) rather than specialized assistive technology to reduce cost, stigma, and maintenance burden.
Relevance
This research demonstrates that accessible exergames can enable sustained independent exercise for people with visual impairments in their homes — addressing a significant health equity gap. The design principles extend well beyond yoga: metaphorical audio instructions, real-time corrective feedback via body tracking, and eyes-free game status communication are applicable to any exercise, rehabilitation, or movement-based application. The finding that mainstream consumer hardware can serve as effective assistive technology reinforces the curb-cut principle and reduces barriers to adoption. For accessibility practitioners, the motivational design techniques (badges, levels, progress tracking conveyed through audio and musical cues) offer a template for making any gamified application accessible. The ABAB single-case experimental methodology is particularly valuable for the accessibility research community, where small participant numbers are common — this rigorous approach allows meaningful evaluation without requiring large sample sizes. The open-source release of the system further supports reproducibility and extension by other researchers.
Tags: visual impairment · exergame · exercise · Kinect · audio feedback · health · motivation · in-home deployment · pose estimation