Android Available – experimental demonstrator Audience: Blind, visually impaired

Owli-AI Way-Buddy

Estimates walkable areas in the camera image and provides acoustic direction cues. An experimental Android demonstrator.

Notice: this page was machine-translated and is pending editorial review.

Good to know

Experimental path estimation can miss obstacles. Distances in the image are not distances in metres. The app does not replace a white cane or guide dog; keep checking your surroundings yourself.

Core features

Walkable path detection

The app analyzes the camera image and estimates which areas might be walkable.
Real-time direction estimation

A recommended movement direction is calculated from the detected walkable area.
Acoustic stereo feedback

A left/right differentiated audio signal provides cues about the estimated direction of movement.
On-device processing

Image analysis runs directly on the smartphone and no camera images are sent to servers.
Training mode for data collection

Optionally, image and mask data can be stored locally to generate training data for model improvements.
Development and analysis mode

Visualizations and debug information provide transparent insight into model predictions.

Privacy

Operating mode: Local

Image processing runs fully on-device. No camera data is transferred to external servers.

Stored training data remains local on your smartphone.

Go to privacy page

Install directly on Android

Install the app from Google Play or scan the QR code if you are viewing this page on a desktop computer.

Install on Google Play View all Owli-AI Android apps

QR code for the Google Play page of Owli-AI Way-Buddy

Open it quickly on your phone

Scan the QR code with your phone to open the store page directly.

System requirements

Android 10 or newer
Camera access

Details and usage

Who is Owli-AI Way-Buddy for?

Owli-AI Way-Buddy is designed for blind and visually impaired people who need additional orientation support while walking outdoors.

How Way-Buddy works

Point the camera
The smartphone is pointed forward so the area in front of you is captured.
Scene segmentation
A trained neural network detects walkable surfaces in the camera image.
Direction calculation
An estimated movement direction is derived from the detected walkable area.
Acoustic feedback
Through headphones or speaker, you receive a left/right differentiated orientation signal.

Training and development mode

Way-Buddy also includes a training mode where image and mask data can be stored locally. This data is used to further develop and improve the underlying models.

A development mode visualizes model predictions directly on screen and makes system behavior transparent.

Privacy and processing

Image processing runs fully on-device. No camera data is transferred to external servers. Stored training data remains local on your smartphone.

Research context

Way-Buddy is not only an app but also an experimental demonstrator within Owli-AI’s assistive computer vision & mobility research strand. Questions around pedestrian-centered scene annotation and segmentation are investigated in the ECCV workshop contribution “Open-Vocabulary Annotation for Pedestrian Scene Understanding under Domain Shift”.