Navigation Pilot
Designing an AI-assisted navigation experience that helps drivers interpret uncertain road risks and make timely decisions — without losing control.
From route guidance to context-aware assistance.
Routes · Traffic · Turn-by-turn guidance
Road signals · Vehicle status · Route context
One whole trip, in one minute.
A 1-minute walkthrough of how Navigation Pilot helps drivers anticipate, assist, and reflect across the trip.
The challenge is not more information. It's knowing what deserves attention.
Drivers already manage routes, road conditions, vehicle status, and attention. Poorly timed assistance can become another source of cognitive load.
When attention is limited
Drivers cannot process every alert at once.
When decisions are time-sensitive
Fuel, parking, and route choices appear while driving.
When feedback lacks meaning
Trip summaries need explanation, not judgment.
How might we design a driving assistant that knows when to warn, assist, reflect — or stay quiet?
Drivers didn't want more alerts. They wanted better timing.
Drivers did not want more alerts. They wanted fewer interruptions, better timing, and explanations they could trust.
TomTom's Virtual Horizon technology already predicts what's on the road ahead. The question was how to present it without adding to the driver's workload.
Road, vehicle, and route context could support decisions before, during, and after a trip.
Trust depends on relevance, timing, and clarity.
Prediction only helps when it arrives at the right time.
The challenge wasn't adding intelligence. It was earning attention.
Four rules the assistant lives by.
- 01
Reduce cognitive load.
An alert that demands attention at the wrong moment is a liability, not a feature.
- 02
Intervene only when necessary.
Silence is the default. Every alert must earn its place in the driver's attention.
- 03
Preserve driver agency.
The system surfaces choices, not commands. Override is always visible and respected.
- 04
Build trust through transparency.
Every action includes a reason. No invisible behavior, no silent handoffs.
Three moments, one quiet assistant.
Predictive Guidance
Warn early, explain clearly, then get out of the way.
Detects upcoming road conditions, explains what to do next, and minimizes after the alert so guidance stays visible without taking over the screen.
Smart Stop Recommendation
Recommend only when the context makes it useful.
Uses vehicle status and route context to recommend useful stops without pulling drivers out of navigation.
Post-Trip Reflection
Turn trip data into understanding, not judgment.
After arrival, a light summary highlights key trip patterns and explains why results changed — so feedback feels like coaching, not grading.
This was a 2024 project. With AI, I'd test the timing, not just the screens.
A working prototype, not frames
Build the alerts as a real, clickable prototype with AI coding tools, instead of a set of static Figma screens.
Test the timing early
Play a drive with alerts firing in real time, and ask drivers which ones came too early, too late, or weren't needed.
Let the AI say why
Prototype the short "reason" line behind every alert with a language model, then test if drivers actually trust it.
Designing AI is not about showing more. It's knowing when to stay quiet.
Navigation Pilot explores how predictive road data, explainable alerts, and restrained feedback can help drivers trust assistance without feeling overwhelmed.