Mission
Independent travel, with an aid that knows its limits
NaviGuider is for blind and low-vision travelers going somewhere unfamiliar, on foot, without a sighted guide. Our job is the part of that walk no existing tool covers well — and to be honest about where that job ends.
Who this is for
NaviGuider is built for independent travelers who already have orientation and mobility training and already use a white cane or a guide dog. It assumes a chest-mounted iPhone so both hands stay free, and it assumes you know how to travel — it is not a substitute for learning how.
It is also useful to a second group: sighted companions, O&M specialists, and developers who need to see what the system perceived and why it said what it said. That is what the live scene is for.
What we are actually trying to fix
Turn-by-turn navigation ends at the block. A cane reads the metre in front of you. Between those two is the part of a walk that makes unfamiliar routes exhausting: where the sidewalk narrows, where the construction hoarding starts, which side of the pole to take, where the curb actually is, whether the crossing is where you expected it.
NaviGuider combines destination routing, short-range path guidance, obstacle and terrain awareness, and conservative crossing assistance into one aid, so that band is covered without juggling four apps in the middle of a street.
What success looks like
Three things, in order:
- The traveler receives one timely, understandable action at a time.
- The system identifies when it lacks the confidence to guide, and says so rather than guessing well.
- An active route stays usable through temporary network loss.
Notice what is not on that list: coverage claims, obstacle-detection percentages, or anything that sounds like autonomy. Those are the numbers that get an assistive product adopted and then get someone hurt.
How we sound
Calm, dependable, respectful. NaviGuider should sound like a skilled guide: concise under pressure, honest about uncertainty, and quiet when there is nothing useful to say. It does not narrate every detected object, gamify safety, or use alarm-heavy feedback. It never says “safe to cross” or “go.”
The five rules we design against
- 01
One action at a time
A traveler crossing an unfamiliar street can act on one instruction. Everything else is noise competing with the sound of traffic. If we have three things to say, we say the one that matters now.
- 02
Uncertainty is a user-facing state
When the app cannot see well enough to guide, that is information the traveler needs immediately — not an internal flag that produces vaguer directions. Low confidence gets said out loud.
- 03
Safety outranks efficiency and polish
A shorter route, a smoother animation, and a more impressive demo all lose to a more conservative instruction. Every time.
- 04
Controls stay stable and eyes-free
Learned muscle memory is worth more than a redesign. Controls stay where they were, operable with VoiceOver, Voice Control, and without sight.
- 05
Visualization explains, never competes
The live scene exists to show why guidance changed. It never takes priority over the spoken or haptic instruction, and it is never required to use the app.
The reasoning behind these is in why we built this, and the boundaries they create are in safety and limitations.