Back to Projects
Assistive Vision & Radar Platform

Sight Bridge AI
Vision Assist

Intelligent companion and walking radar for the visually impaired. Combining ultra-fast On-Device computer vision with Cloud Multimodal Intelligence and smart emergency SOS safety.

Get it on Google Play
On-Device LiteRT & Gemma
Bilingual TTS (VI & EN)
Sight Bridge AI Logo

Blind-Friendly Gesture Touchpad

Engineered with high-contrast Safety Yellow UI and full-screen tactile gesture mappings for effortless eyes-free navigation.

Step 1

Single Tap

Repeat last announcement / Cancel active countdown

Step 2

Double Tap

Instant panoramic hazard scan

Step 3

Triple Tap

Open system audio & sensor configurations

Step 4

Press & Hold

Speak to Multimodal AI Assistant

Engineered for Safety & Independence

A comprehensive safety architecture combining offline real-time edge intelligence with multimodal generative AI.

Sub-25ms Reflex

Walking Safety Radar (100% On-Device)

Ultra-fast sub-25ms reflex processing powered by on-device computer vision models running 100% offline. Detects parked motorbikes, pedestrians, stairs, drop-offs, and curbs with directional haptic pulse and voice alerts.

Gemini & Gemma AI

Multimodal Push-to-Talk Assistant

Press and hold anywhere on the screen to query your surroundings: scene description, sign reading, or spatial analysis powered by Google Gemini & Gemma with natural voice responses.

GPS Auto SOS

Emergency SOS & Fall Detection

Precision accelerometer algorithms detect sudden falls and impacts, automatically triggering a 10-second countdown before dispatching emergency SMS with accurate Google Maps GPS coordinates.

Blind-Friendly

Accessible Full-Screen Touchpad

Intuitive blind-friendly gesture system (Single Tap: repeat/cancel, Double Tap: scene scan, Triple Tap: settings, Hold: talk to AI) with high-contrast Safety Yellow on Pure Black UI.

EN & VI Support

Full Multilingual Engine

Seamless localization across HUDs and assistive features, automatically switching between native US English (en-US) and natural Vietnamese TTS speech synthesis.

Zero Data Logging

Privacy-First & BYO Gemini API Key

Camera frames are processed in real-time and never harvested. Run 100% off-grid with LiteRT/Gemma or plug in your personal Google Gemini API Key for cutting-edge cloud models.

Privacy & Security Commitment

Your vision data stays strictly yours.

Camera streams are processed live in device volatile memory and never recorded or stored. If you connect your Google Gemini API Key, it is stored in encrypted on-device keystore, ensuring complete data sovereignty.

100% Offline Radar Operation
Encrypted Local Key Storage
Get it on Google Play