Sound Navigation Demo

Semantic and Spatial Sound-Object Recognition for Assistive Navigation

Panning Camera

Distance Camera

Interaction Level:
Objects are grouped into high, medium, or no interaction based on usefulness.

High are important for navigation.

Medium are somewhat useful.

None are not useful.

This is represented by sound volume changes using simple tones and can be personalized for every user.

Obstacle Width to Pitch Mapping:

The width of an object is represented by the pitch of its sound.

We use three discrete categories: narrow, medium, and wide.

Medium is the baseline pitch.

Narrow objects sound one octave higher.

Wide objects sound one octave lower.