UX case study · Accessibility · Published HMI research
Autonomous Rideshare
An independent ride should not depend on seeing the screen.
A multimodal human-machine interface for autonomous vehicles, tested in a controlled study using simulated visual impairment and published by SAGE in 2025.
- Problem
- Vision-first vehicle interfaces leave blind and low-vision riders without accessible trip information.
- My role
- UX researcher and publication co-author: literature review, protocol design, study sessions and analysis.
- Timeline
- 2.5 months
Published by SAGE, 2025 - Outcome so far
- In an N=24 simulation study, audio narrowed the gap with sighted participants; visual-only performed significantly worse (p < .05).
01 · Context & stakes
Most autonomous HMIs are vision-first.
Autonomous vehicles promise independent mobility, but blind and low-vision riders face barriers at every stage: identity verification, trip confirmation, en-route updates, unexpected events and safe exit.
The stakes
25 million Americans
Face transportation insufficiency related to sensory, cognitive or motor impairments. A trip is only independent if the rider can understand and act on the information it provides.
The design question
Can another channel close the gap?
We investigated whether synchronized audio and visual feedback could improve trust, vehicle identification and navigation when visual information becomes inaccessible.
02 · Role & boundaries
What I owned
My contribution combined hands-on research with shared responsibility for the study protocol, analysis and manuscript.
I drove
- Conducted the 27-paper literature review.
- Co-designed the protocol: three conditions and six tasks.
- Conducted 8 of 24 Wizard-of-Oz sessions and post-session interviews.
- Co-authored the codebook and coded 8 of 24 transcripts.
- Contributed to Kruskal-Wallis interpretation and co-authored the Methods and Results sections.
Others contributed
- The research team shared protocol design and study execution.
- Transcript coding discrepancies were resolved by consensus.
- Co-authors shared analysis and manuscript development.
Constraints
- 2.5-month research timeline.
- 24 participants across three conditions.
- A controlled Wizard-of-Oz environment, rather than an on-road deployment.
- Simulated 20/200 acuity; lived experience of visual impairment remains a validation gap.
03 · Research
27 articles. One testable gap.
We bounded the literature review at 2014, filtered by keyword and abstract, and read 27 articles in full. Few designs used a multimodal approach. That gap became a controlled comparison of the same journey with and without audio.
Literature review
Accessible HMI and rideshare
Search terms covered accessible HMI, accessible rideshare, visually impaired HMI, fully autonomous vehicles and multimodal design.
Six journey tasks
From verification to exit
Each condition covered the same six trip tasks, with measures of trust, satisfaction and navigation accuracy.
Mixed evidence
Ratings and interviews
Post-session interviews complemented the task measures. We used a shared codebook and resolved coding discrepancies by consensus.
Between-subjects. Three conditions.
Twenty-four participants were randomized across a sighted multimodal baseline, simulated visual impairment with audio, and simulated visual impairment without audio.
Non-visually impaired
Sighted participants, full multimodal interface. The baseline.
Simulated impairment + audio
Cambridge simulator, 20/200 acuity, full multimodal interface.
Simulated impairment, no audio
Visual-only control — isolating audio's contribution.
What this method can establish: the Cambridge Disability Simulator standardized sensory degradation across conditions. It does not reproduce the adaptive expertise or lived experience of blind and low-vision riders.
04 · The solution
One ride. Three connected components.
A rear-seat display, synchronized voice prompts and tactile controls support the rider through the same six tasks.
Display
Rear-seat display
High-contrast route status and environment descriptions, synchronized with audio for redundancy.
Voice
“Maple”
A natural, calm voice character delivering synchronized, proactive prompts across all six tasks.
Tactile
Hardware buttons
Two high-contrast, color-coded tactile buttons allow interaction without a touchscreen.
Wizard-of-Oz simulation replicating the rear-seat AV experience with synchronized audio and visual output.
Full system mockup — the proposed in-vehicle concept.
Six tasks, end to end
We mapped the full rideshare journey with impairment needs included, then translated each stage into an accessible interaction to evaluate.
Identity verification
A spoken 4-digit code replaces visual QR codes blind riders can't use.
Trip confirmation
Destination read back aloud; the rider verbally confirms or corrects.
Driving updates
Proactive narration — intersections, signals, ETA — keeps riders oriented.
Unexpected events
A special mode explains hazards and evasive actions — trust when the vehicle deviates.
Destination arrival
A 3-minute pre-arrival alert with curbside context and orientation.
Exit interaction
Step-by-step guidance intended to support safer, better-oriented disembarkation.
05 · Key decisions & trade-offs
Give the rider information they can use.
Three design principles connected the accessibility gap to concrete interaction choices.
Memory supports
Make the next step explicit
Spoken verification, destination readback and a three-minute arrival alert make information available at the point of action. Predictive cues and consistent language reduce what riders must remember.
Trade-off: proactive prompts add information throughout the trip. Rider control over pace and cue frequency remains a next step.
Perception optimization
Repeat across channels
Audio and visual feedback carry the same trip information; tactile buttons provide a way to act without the touchscreen. Progressive disclosure supports access as the journey changes.
Boundary: this study isolated audio's contribution. Combined sensory impairments need further work on haptics and localized speakers.
Attention management
Match cues to urgency
Routine navigation updates orient the rider. A special mode explains hazards and evasive actions when the vehicle deviates, making high-priority information salient.
Trade-off: keeping riders informed must be balanced against overwhelming them. Personalization of verbosity and tone remains to be tested.
06 · Outcome & evidence
Audio narrowed the gap.
In the simulated visual-impairment condition, multimodal feedback brought vehicle identification and navigation toward the sighted baseline. Visual-only performed significantly worse (p < .05). Audio narrowed the trust gap; it did not fully close it.
Study scope
N=24
Three conditions and six trip tasks in a controlled simulation.
Reported difference
p < .05
Visual-only performed significantly worse than multimodal.
Hypotheses
3 of 3 supported
Navigation, near-parity and the effect of removing audio.
H1 supported — audio improved navigation confidence and route awareness.
H2 supported — multimodal feedback enabled near-parity with sighted riders.
H3 supported — removing audio decreased trust and situational awareness.
The full picture: multimodal audio-visual feedback improved route awareness, destination confidence, vehicle identification, and satisfaction versus visual-only. Honest nuance: audio narrowed the trust gap with sighted riders — it didn't fully close it.
SAGE · Peer-reviewed · 2025
The study was published as research evidence for multimodal autonomous-vehicle interfaces. These findings describe the controlled study, not a deployed service.
Read the full paper →07 · Reflection
“Did you test with blind and low-vision riders?”
The study used simulated impairment. That made sensory degradation consistent across conditions, but it cannot stand in for lived experience. The evidence supports audio's contribution in this setup; the highest-priority next step is research with actual blind and low-vision riders.
Where this research goes next
Field trials with blind and low-vision riders
Real AVs, varied road conditions — beyond the simulation environment. The highest priority.
Tri-modal redundancy
Seat-vibration haptics and localized speakers for riders with combined sensory impairments.
Personalization
Rider control over verbosity, pace, tone, and cue frequency — adapting to needs and anxiety levels.
A 2x2 replication
Impairment x audio condition — isolating whether sighted riders also benefit, strengthening the causal claim.