UX RESEARCH · INTERACTION DESIGN · ACADEMIC PROJECT

Designing a Clearer Autonomous Ride

Academic project using publicly available Waymo footage; not affiliated with Waymo.

My Role

UX Research
UI & Motion Design
Eye-Tracking Study & Analysis

Team & Duration

3 designers
2 engineers
11 months

The Problem

Unexpected stops and route changes had limited or inconsistently presented explanations.

The Question

How might we help passengers find and understand explanations for unexpected vehicle behavior?

Our Answer

I placed contextual explanations and animated status icons near the vehicle icon to connect its behavior with the reason behind it.

PROBLEM

Could passengers find the explanation when the vehicle behaved unexpectedly?

Upper-screen text display

In the footage we studied, some explanations appeared as text near the top of the passenger display, while other situations offered limited information about why the vehicle was stopping or changing its route.

We wanted to understand whether participants noticed these messages and how they interpreted the vehicle’s behavior.

Original Waymo upper-screen text display

SOLUTION

Bringing vehicle-status explanations closer to the vehicle icon

We designed contextual callouts that pair short explanations with animated icons near the vehicle icon. The design aimed to make the reason for a stop or route change easier to locate.

UI Structure

1. Pair short explanations with visual cues

Short explanations paired with visual cues

2. Place explanations near the vehicle icon

Pilot observations showed that participants often looked near the vehicle icon while giving limited attention to top-aligned status text. We placed callouts above the icon to connect each explanation visually with the vehicle.

Vehicle-anchored pedestrian detection prompt

Pilot Test

Exploring how participants noticed and interpreted vehicle-status messages

We used Tobii eye tracking while participants watched selected autonomous-driving footage, then gathered questionnaire and interview responses to understand what they noticed and how they interpreted the vehicle’s actions.

Working with a professor and three student teammates, I helped design the study, conducted sessions and interviews, analyzed interview responses, and contributed to interpreting gaze data.

Evaluating Passenger Anxiety Based on Interface Feedback in Autonomous Driving Emergencies

Team A

Video Viewing

Eye Tracking

AS-IS

Post-Questionnaire

Team B

Video Viewing

Eye Tracking

TO-BE

Post-Questionnaire

Comparison of Anxiety Metrics

Fixation Duration

Quantitative Survey Results

Final Insights

The improved interface reduced passenger anxiety during sudden autonomous driving events, such as emergency braking, as indicated by shorter eye-tracking fixation durations.

Hwang, S., Kim, N., & Lee, D. (2023). A study on passenger anxiety evaluation in adverse weather conditions for autonomous driving from the user perspective. Journal of the Korean Society of Transportation, 41(1), 104–118.104-118.

https://doi.org/10.7470/jkst.2023.41.1.104

AS-IS

Post-survey

Comparison of anxiety indicators by TOI

fixation duration by video scenario

Conclusion drawing

Collected gaze data via Tobii during autonomous taxi video viewing and analyzed gaze patterns with post-surveys.

Conducted an experiment with five male and female participants aged 20–40, encouraging open discussion with a moderator acting as a co-passenger.

Qualitative feedback from participants was analyzed using affinity mapping.

Limited contexts for

information delivery

Various detailed situations are visually communicated using custom icons.

Information delivery

location and format

The information is conveyed via a speech bubble positioned just above the vehicle icon.

Problem

Solution

FROM INSIGHTS TO DESIGN

Translating pilot observations into design decisions

The pilot highlighted two issues: limited explanations in some situations and a mismatch between message placement and visual attention.

Limited explanations during unexpected behavior

Participants described uncertainty when the vehicle appeared to change its route without a clear explanation.

→ Design response

Add messages that describe the relevant situation or vehicle action.

[Scenario 1: Route Confusion]

[Scenario 2: Route Confusion]

Status text received limited attention

During an unexpected braking scenario, participants gave little attention to the top-aligned explanation while looking near the vehicle icon.

→ Design response

Place contextual callouts near the vehicle icon, where participants were already looking.

[No Visual Attention to Top-Aligned Text]

INTERFACE DESIGN | Contextual Callouts

Short messages paired with visual cues

The interface combines brief status messages, icons, and motion cues in callouts above the vehicle icon. The visual system includes light and dark variants.

Prompt UI : Light mode

Light mode

Rerouting

Spot to pullover

Person ahead

Busy intersection & Pedestrian detected

Busy intersection

Prompt UI : Dark mode

Dark mode

Rerouting

Spot to pullover

Vehicle on the left

Pedestrian detected & Busy intersection

Pedestrian detected

FINAL DESIGN

From top-screen text to vehicle-anchored prompts

We moved explanations closer to the vehicle icon, drawing on the attention patterns observed in the pilot.

AS-IS

Top-aligned status text

In the selected footage, status explanations appeared near the top of the display, separate from the vehicle icon.

Original interface with explanations at the top of the screen

TO-BE

Callouts near the vehicle icon

The redesign placed contextual explanations above the vehicle icon and paired them with visual cues.

Redesigned interface with a prompt above the vehicle icon

EVALUATION & FINDINGS

Comparing the redesign with the baseline

We conducted a screen-based, between-subjects comparison across eight driving scenarios. Of 23 recruited participants, 22 were included in the analysis: 11 viewed the baseline interface and 11 viewed the redesign.

We examined gaze patterns, situation recognition, perceived ease of finding information, satisfaction, and trust, alongside post-study interviews.

FINDING 01

Recognition varied by scenario

The redesigned group had more correct recognition responses in some scenarios and fewer in others. These descriptive results do not establish a consistent or statistically significant improvement.

FINDING 02

Satisfaction and trust showed no significant differences

Overall satisfaction and trust did not differ significantly between the two conditions. Although average trust was lower in the redesigned group, the study did not establish a decrease in trust.

Points show group means; error bars show 95% confidence intervals. Both measures used a 7-point scale, with 11 participants per condition.

Satisfaction and trust showed no significant differences — comparison chart

LESSON I LEARNED

Distinguishing attention, understanding, and trust

Use observation to guide design, then test the assumption

The pilot showed where participants looked and helped us decide where to place explanations. The follow-up study reminded me that attracting attention does not, by itself, establish better understanding or greater trust.

Combine behavioral and self-reported evidence

Eye tracking showed where participants looked, while interviews and questionnaires helped us examine how they interpreted and felt about the experience. These methods addressed different questions and needed to be interpreted together.

What I would test next

I would vary placement, wording, and motion separately, and assess understanding immediately after each scenario to reduce reliance on later recall.