Quantifying the Effect of Visual Impairments on Daily Household Activities in Virtual, Interactive Environments
Wensi Ai, Sharon Lee, Li Fei-Fei, Jiajun Wu, Ruohan Zhang
Cog Sci 2023 | Paper
Contributions
A novel approach to quantify the impact of visual impairments (VIs) on daily household tasks
Developed visual field models in virtual reality (VR) with clinical data from VI patients
Leverage the state-of-the-art simulation environment to recreate six common daily activities
Model how body and eye movement strategies affect task performance under the influence of VIs
Video Demos
Experiment Design
Catch
Navigate
Place
Slice
Throw
Wipe
3-layer rendering pipeline to simulate how light passes through the human eye
6 household tasks to capture essential visuomotor skills required for daily household activities
Questionnaires empathy questionnaire and self-reported difficulty ratings from subjects.
Movement metrics motion and eye data collection used to examine strategies used by subjects to overcome different tasks under different VI types and conditions
Visual Impairment Simulation Pipeline
Results
The above figure shows the Performance Drop Ratio compared to the normal condition of each task, averaged across subjects. Late-stage glaucoma and cataracts significantly reduce all 6 tasks performances.
E.g. the navigate task within late-stage glaucoma represented by the top magenta triangle shows that the subjects' performance dropped approximately 20 times in late stage glaucoma compared to the normal baseline (normal vision). See Remarks in the Results section.
The above figure shows subject difficulty ratings for each VI condition compared to normal vision.
E.g. a higher rating implies a higher impact of VI, where 1 represents "no impact" and 5 represents "extreme impact".
Key findings: Place with AMD and glaucoma was rated the most difficult; this is followed by AMD, cataracts, myopia, and presbyopia in the respective order.
The image carousel shows averaged dependent variables aggregated across all tasks. The table highlights the dependent variables (listed below) in each task-VI combination where a significantly low score is represented in red and a significantly high score is represented in blue.
More information on the results can be found in the 'Results' section under 'Effect of Movement on Task Performance' in the paper submission.
Regression Model Results, each cell in the format of [dv(coeff, p)] which represents the significant dependent variables (listed below), and their coefficients in the model with its p-value. The dependent variables listed are as follow:
gm: gaze movement velocity, bt: body translation velocity, br: body rotation velocity, ht: head translation velocity, hr: head rotation velocity, rt: right-hand translation velocity, rr: right-hand rotation velocity, lt: left-hand translation velocity, lr: left-hand rotation velocity
More information on the results can be found in the 'Results' section under 'Effect of Movement on Task Performance' in the paper submission.
Additional Results: Empathy Questionnaire
Previous studies has shown that immersive experiences of VI studies in VR fostered empathy among subjects. Here, we constructed a 22-question survey to measure empathy and attitudes toward VI patients. Our results show that our study fostered subjects' empathy towards VI patients.
More information on the results can be found in the 'Results' section under 'Empathy Study Results' in the paper submission.
Empathy Results 17 out of 22 survey items show an increase in the average attitude/ empathy score across all subjects compared to their scores before the VR study. The increase is statistically significant for Q1.