Obstacle Detection in Crowded Spaces

Moving through a station concourse at rush hour without a guiding arm.
Crowds are the hardest case: the obstacles move, they move unpredictably, and a cane finds them a step too late. We ran Phoenix Eye through four weeks of peak-hour testing in a mainline station, tuning how early a warning arrives and — just as importantly — when the glasses stay quiet. Constant narration is worse than none. The device now speaks only when something on your path changes: a suitcase left standing, a person cutting across, a queue that has formed since you last passed.

Delivered
Obstacle model tuned for dense pedestrian traffic, with warning distance and alert frequency set from tester feedback.


