Challenge
Drivers need information without being overloaded by it.
Modern in-car interfaces expand what drivers can do, but the same interface can become distracting when attention needs to stay on the road.
NOVA
An adaptive automotive HMI that changes with the driver’s level of attention.
Nova rethinks how information, interaction, and automation should shift as responsibility moves between the driver and the vehicle.
Quick read
Approach
Make the interface respond to driving context.
We structured information and interaction around changing levels of driver attention and vehicle automation.
Outcome
A three-mode adaptive HMI, validated through iterative testing.
The final prototype reached a SUS score of 91.2 and was refined through in-person, remote, and A/B testing.
The problem
The more drivers can do, the more attention the interface can demand.
In-car screens have become essential for navigation, media, communication, and vehicle control. But while driving, every added interaction competes for a limited resource: attention.
4.8-13.6 sec
Complex touchscreen interactions can divert a driver’s attention for several seconds at a time.
Problem Visual
How much should the interface ask from the driver at any given moment?
Distraction isn’t one thing.
Driving interfaces can compete with attention visually, manually, and cognitively. The risk increases when a single interaction demands all three at once.
VisualEyes leave the road
+ManualHands leave the controls
+CognitiveAttention leaves the driving task
↓
High interaction cost
Design implicationInstead of treating every feature as equally available, we needed to control when information and interaction should surface.
I wanted to understand what drivers actually reach for, and when.
As project lead, I guided our research around everyday driving behavior, interface preference, and attitudes toward assisted driving.
Research Photography
- Photo DiariesEveryday controls and interfaces
- Surveys + InterviewsBehavior and comfort around device use
- In-Vehicle ObservationHands-on exploration of a modern EV interface
Synthesis
The interface needed to know when to step forward, and when to get out of the way.
- 01
Clarity before capability
Drivers need to understand what the vehicle is doing before they can trust what it can do.
- 02
Attention changes with context
A dense interface may be useful when attention is available and dangerous when it isn’t.
- 03
Automation changes what the interface can become
As driving responsibility shifts, previously restricted information and interactions can safely surface.
How might the HMI adapt as the driver’s attention changes?
System architecture
One HMI. Three levels of attention.
Driver attention
HighLow
- Focus
- Comfort
- Relax
- Driving responsibility
- Driver-ledAssisted / sharedSignificantly reduced
- Driver attention
- HighSharedLow
- Information density
- MinimalBalancedExpanded
- Interaction depth
- Essential onlyContextualRicher / immersive
Focus
Driver-led. Driver attention: High. Information density: Minimal. Interaction depth: Essential only.
Comfort
Assisted / shared. Driver attention: Shared. Information density: Balanced. Interaction depth: Contextual.
Relax
Significantly reduced. Driver attention: Low. Information density: Expanded. Interaction depth: Richer / immersive.
Driver attention
HighLow
Driver attention
HighLowFocus
Driver-led
- Driver attention
- High
- Information density
- Minimal
- Interaction depth
- Essential only
Comfort
Assisted / shared
- Driver attention
- Shared
- Information density
- Balanced
- Interaction depth
- Contextual
Relax
Significantly reduced
- Driver attention
- Low
- Information density
- Expanded
- Interaction depth
- Richer / immersive
Before designing screens, we defined what deserved attention.
I structured the HMI around information priority rather than individual screens, defining what should remain persistent, what could appear contextually, and what should stay hidden until the driving state allowed it.
Information priorityFocusComfortRelax
Layer 01
Persistent
Speed · Vehicle status · Driving mode · Critical warningsPrimary
Persistent
Persistent
Layer 02
Contextual
Navigation · Cameras · Notifications · ScheduleHeld back
Selected
Contextual
Layer 03
Expanded
Media · Communication · Entertainment · Environmental informationHeld back
Held back
Expanded
Layer 01
Persistent
Speed · Vehicle status · Driving mode · Critical warningsLayer 02
Contextual
Navigation · Cameras · Notifications · ScheduleLayer 03
Expanded
Media · Communication · Entertainment · Environmental informationPrimarily Layer 01
Layer 01 + selected Layer 02
Layer 01 + Layer 02 + expanded Layer 03
The system scales information density with the amount of attention available.
Structure first. Visual detail second.
Once the behavioral rules were defined, we translated the architecture into progressively higher-fidelity prototypes, using each stage to refine hierarchy, interaction, and visual emphasis.
Lo-Fi
Lo-Fi
Information zones
Mid-Fi
Mid-Fi
Hierarchy and interaction
Hi-Fi
Hi-Fi
Visual language and state feedback
Testing showed where calm became unclear.
Key testing themes
- 01
overly dominant speedometer presentation
- 02
small or unclear interactive controls
- 03
related actions being visually separated
- 04
limited personalization
- 05
mode transitions needing clearer feedback
- 06
layouts benefiting from further simplification
Before
Related actions were visually separated.
Before
After
Grouped contextual actions reduced search effort.
After
Final experience
Focus
Only what matters when the driver is in control.
Focus prioritizes essential driving information and suppresses unnecessary interaction for dense, attention-heavy environments.
Focus
Comfort
More context when the vehicle can share the workload.
As assistance increases, Comfort introduces contextual information without removing the driver from the loop.
Comfort
Relax
When attention is available, the interface opens up.
Relax expands the experience toward media, environment, and shared content when driving responsibility is significantly reduced.
An interface designed around attention, not screen space.
91.2 System Usability Scale
Testing reinforced a simple principle: clearer hierarchy and more deliberate transitions helped the interface feel calmer and easier to understand.
What I’d carry into the next HMI I design
Nova changed how I think about automotive interfaces. The hardest problem wasn’t deciding what information a vehicle could show, but deciding what deserved the driver’s attention at any given moment.
Leading the project pushed me to think beyond individual screens and design relationships between context, responsibility, information, and interaction. That systems-level thinking is something I want to continue exploring in automotive HMI and other complex, safety-critical experiences.