An autonomous robotaxi cabin that earns passenger trust through empathy, not explanation.
Period
2024
Type
Interaction Design · Cabin HMI
Status
Course project · team of two
Role
Topic selection, concept definition & design, persona analysis, user journey analysis, information architecture design and HMI design
The car is not a machine that drives you. It is a companion that tries.
Trust Is the Design Problem
Passengers of autonomous vehicles receive too little information about the vehicle's
surroundings and performance. This information asymmetry makes them anxious: they worry the
system cannot respond correctly in complex situations, and they cannot empathize with—or
trust—a car they do not understand.
Context-adaptive HMI concepts can close part of the gap: providing situation-specific
information helps passengers understand the vehicle's decisions and builds trust. The design
opportunity is to go further—to make the vehicle's effort legible as emotion.
Competitor analysis
Waymo One
Transparent operations
Transparent trip information, controllable cabin settings, emergency support—but fixed
routes and little emotional care.
Control + safety 萝卜快跑 Apollo Go
Mature voice, low transparency
Mature multimodal voice interaction and rich entertainment, yet limited emotional
interaction and low decision transparency.
Voice + media Zoox
Surrounding social cabin
Surrounding seats promote interaction; light and ambience build emotional connection, but
the single-purpose design limits long solo rides.
Social space Tesla Robotaxi
Efficiency over care
Deeply integrated ecosystem and strong cost control, but weak emotional support and
comfort.
Ecosystem 极氪 001 ZEEKR
Therapy as a differentiator
Immersive healing modes combine light, sound and touch with health monitoring—strong for
anxiety relief, narrow in application.
Healing mode
Questionnaire
A survey asked different groups about their experience with today's taxis and their view of
driverless robotaxis.
Respondents by age
50%18–30
36%30–45
14%45+
Current taxi satisfaction
18%Satisfied
58%Neutral
24%Dissatisfied
What passengers care about
Top concern
Safety
Can the vehicle guarantee a safe ride?
Security Top concern
Punctuality
Will I still arrive on time?
Time Top concern
Price
Is the ride worth the cost?
Cost Top concern
Comfort
Can I relax during the ride?
Comfort
Willingness
Roughly half of the respondents were willing to try a robotaxi; the rest were interested but
cautious, uncertain, or unwilling.
Persona: Zhang Hong
User Positioning
Zhang Hong, male, 32, Shanghai — product manager at a multinational company.
User positioning
A young user willing to try new things. He commutes by taxi, mainly for commuting and
business; his schedule is tight and he has a strong sense of time; his rides are long and
often caught in traffic.
Riding Habits
He frequently travels for work, often leaving during the evening rush to protect his
daytime work. International trips take him to the distant Pudong Airport. He has strong
emotional control in the workplace — but on taxi rides to the airport, unexpected
situations regularly make him lose control.
Needs and Tasks
Functional needs
01
Smart routing
Plan routes intelligently to avoid congested sections.
Route 02
Ease the irritation
Help relieve the passenger's mood during jams.
Emotion 03
Show safety status
Display the autonomous vehicle's safety state.
Safety 04
Transparent decisions
Make the decision process visible.
Trust
Task analysis
He takes a taxi to the airport; the ride encounters jams and dangerous conditions; his
emotions shift with the driving situation.
Goals
User goals
Behavior
Arrive on time
Take the taxi and reach the airport punctually.
On time Experience
Arrive without drama
Arrive at the airport without scares — and feel that the taxi empathizes with him and
is trying.
Empathy Long-term
Every trip pays off
Business trips get rewarded; every journey feels worthwhile.
Value
Business goals
01
Rebuild trust
Reduce distrust of autonomous vehicles and expand the customer base.
Growth 02
Keep passengers
Arrive on time, raise ratings and improve retention.
Retention 03
Guarantee safety
Ensure passenger safety and reduce loss.
Safety
The Key-Path Journey
The journey in three stages
Stage 1 · Boarding
Trusting the car to drive
Behavior: confirms the route and destination; watches the car drive; does non-driving
tasks. Emotion: anticipation. Trust: doubts whether the autonomous vehicle can drive
safely.
Doubt Stage 2 · Mid-jam
Trusting the car to be on time
Behavior: keeps monitoring the traffic; judges the cause and expected delay; replans
the schedule; tries alternate routes. Emotion: disappointment, doubt, anxiety. Trust:
doubts on-time arrival; cannot feel the car's empathy.
Anxiety Stage 3 · After the jam
Trusting the car to be trying
Emotion: anxiety, half-belief. Trust: still doubts whether the car understands the
need and is driving hard enough.
Half-belief
Pain points
Anxiety and helplessness from the risk of delay—before the journey starts and while the
jam is happening.
Design opportunities
01
Strengthen situational awareness
Optimize the information given to the passenger.
02
Add emotional interaction
Build trust and relieve anxiety through car–passenger emotional interaction.
03
Make the effort legible
Use light, sound and other multimodal cues to express—even exaggerate—the car's
driving state, increasing empathy-trust.
Information Architecture
The cabin information is layered by urgency. The first layer: speed, basic safety, arrival
time, next maneuver, destination distance. The second layer: more detailed safety and route
information. The third layer: internal component states, destination recommendations and
coordination with the dispatch backend.
Information Architecture & Layout
Emotional interaction system
01
Virtual persona
Voice and touch interaction with the virtual persona.
Dialogue 02
Real-time emotion monitoring
Monitors the passenger's emotional state in real time.
Sense 03
Task-based empathy
Empathy understanding built on the passenger's task and context.
Understand
Three layers of information
Layer 1
Core status
Speed, basic safety and perception of the surroundings, whether the ride is safe,
battery level; arrival time, next maneuver, destination distance, whether the
arrival is on time.
At a glance Layer 2
Detail on demand
More detailed safety information such as exterior damage; route choices and
per-section driving conditions; destination status and recommendations.
On request Layer 3
Internal state
Component-level states such as battery temperature and tire pressure.
Deep system
An Empathetic Vehicle
The core move is to treat the vehicle as a living companion rather than a control panel.
Passengers tap the touchable leather to wake the car's persona, talk to it, and read its state
through a minimal kaomoji-like face and ambient light across the windshield.
Interaction Design
01
Pat
Tap the touchable leather to wake the robotaxi's virtual persona.
02
Stroke
“Shh — this is a little tool we will use later.”
03
Voice
Talk with the virtual persona through voice interaction.
04
Face & Light
Convey the vehicle's state to the passenger through expressions and light.
05
Console
Most complex information still flows through the central console.
The Vehicle Persona
Pet-like interaction design.
01
Kaomoji-like minimal expressions
The simplest combinations of graphic symbols convey the persona's expressions. A
borderless design turns the entire windshield into the face, so passengers feel it
is the vehicle itself talking—not some other character.
02
Natural tactile interaction
Embodied touch enters the human–vehicle relationship, mimicking how people naturally
interact with pets, building closeness and strengthening trust in the vehicle.
03
Bionic natural response
When the passenger interacts by touch, the vehicle's expressions, sounds and lights
respond subtly—simulating emotional exchange between a person and a pet.
04
Gamified interaction in specific scenarios
In specific scenarios, tactile play with the vehicle improves the passenger's mood,
deepens the human–vehicle bond and strengthens empathy-trust in the vehicle.
Validation Scenarios
The concept is walked through six scenarios on the journey to the airport: normal driving;
congestion I (the car explains the accident and replans the route); congestion II (the car
detects anxiety, plays music and turns the wait into a touch game of clearing obstacles);
congestion III (the car announces it has passed the jam and will keep trying); and urgent
driving I–II (when the passenger urges speed, the car changes its lights and plays an engine
sound to show it is trying).
Normal Driving
Overview
The passenger boards; the vehicle starts driving under normal road conditions.
Example
The car
Welcomes and reassures
“Welcome, passenger ending in 1234. Destination: Shanghai Hongqiao International
Airport. Estimated arrival in 24 minutes — 30 km in total.” · (The passenger taps the
phone on the NFC pad) Asks for the last digits of the phone number. · “I know you are
taking flight CA0001 to Beijing. I will dynamically replan the route based on your
flight. Trust me — I will get you there on time.” · (In an emergency) “I am about to
make an urgent turn to avoid danger. Please hold on.”
Car behavior The passenger
Boards and holds on
Boards. · States the destination. · (Holds on.)
Passenger behavior
Congestion I — The Car Explains
Overview
The vehicle begins to enter a congested section; the passenger notices the jam and asks
what is happening.
Example
The car
Explains and promises
(The passenger pats and asks what is wrong) Explains the situation — “There is a
sudden accident ahead; the passage time is expected to be about 20 minutes. Trust me —
I will dynamically replan the route to get you there on time.”
Car behavior The passenger
Asks what is happening
Asks what is happening.
Passenger behavior
Congestion II — Play to Clear Obstacles
Overview
The passenger interacts with the car through gamified interaction — the car detects
anxiety, plays music to calm the mood, and begins the game.
Example
The car
Monitors and plays
Detects the passenger's emotional change. · Plays music to ease the mood. · Engages in
gamified interaction — “Don't worry, trust me, I will get you there on time.” · “The
accident ahead is being handled — stroke the panel and clear the obstacles with me!
When they are cleared, the road ahead will open.”
Car behavior The passenger
Plays along
Plays with the car — strokes the leather trim; the mound of soil beside the face is
swept away with the passenger's movement.
Passenger behavior
Congestion II · Continued
Overview
The passenger interacts with the car through gamified interaction — the car detects
anxiety, plays music to calm the mood, and begins the game.
Example
The car
Monitors and plays
Detects the passenger's emotional change. · Plays music to ease the mood. · Engages in
gamified interaction — “Don't worry, trust me, I will get you there on time.” · “The
accident ahead is being handled — stroke the panel and clear the obstacles with me!
When they are cleared, the road ahead will open.”
Car behavior The passenger
Plays along
Plays with the car — strokes the leather trim; the mound of soil beside the face is
swept away with the passenger's movement.
Passenger behavior
Congestion III — Leaving the Jam
Overview
Leaving the congested section.
Example
The car
Talks and reassures
Talks with the passenger — “We have passed the congested section. I will keep trying
to get you there on time.”
Car behavior
Urgent Driving I — Showing the Effort
Overview
The passenger feels time may not be enough and urges the vehicle to drive harder.
Example
The car
Shows the effort
After receiving the request or detecting the passenger's state — changes the lights ·
plays the engine sound.
Car behavior
Urgent Driving II · Continued
Overview
The passenger feels time may not be enough and urges the vehicle to drive harder.
Example
The car
Shows the effort
After receiving the request or detecting the passenger's state — changes the lights ·
plays the engine sound.
Car behavior
Evaluation
The design was evaluated through a cognitive walkthrough, followed by usability tests with
trust, SART, SUS and acceptance scales.
Cognitive walkthrough.Cognitive walkthrough.
Results
Compared with a lifeless HMI, the empathetic cabin reduces the abruptness of situational
changes, keeps passengers alert and improves situational awareness (SART). Passengers find
the vehicle's behavior easier to understand and trust its information and decisions (trust
scale), although some feel less familiar with the more complex system. Usability improves,
especially independent use without professional help (SUS), and acceptance is
higher—passengers form more positive emotions, find the vehicle more effective, and see a
real need for the system.
Usability test session.Usability test session.
What this is
A cabin HMI that treats the autonomous vehicle as an empathetic companion—making its effort
feelable—and validates that this rebuilds passenger trust.
The Empathetic Robotaxi was designed with Li Tinghao for the Interaction Design course,
supervised by Prof. You Fang.
