Skip to content

JoshuaChil/EdgeVision

Repository files navigation

EdgeVis

EdgeVis is a mobile AI navigation assistant that runs entirely on-device, using Vision-Language Models (VLMs) integrated with the RunAnywhere SDK in a React Native application.

Unlike cloud-based assistants, EdgeVis performs perception, reasoning, and navigation locally on the device—enabling low-latency, privacy-preserving, and offline-capable AI navigation.


🚀 Overview

EdgeVis combines real-time visual understanding with natural language reasoning to help users navigate their environment without relying on cloud infrastructure.

By leveraging on-device VLMs and the RunAnywhere SDK, EdgeVis delivers consistent AI-assisted navigation even in offline or low-connectivity environments.


🧠 Key Features

  • Fully On-Device AI

    • Vision and language inference run locally
    • No cloud calls or remote processing
  • Vision-Language Navigation

    • Interprets camera input and user queries
    • Provides contextual navigation guidance
  • Privacy-First Architecture

    • Images and sensor data never leave the device
    • Suitable for sensitive or regulated environments
  • Low Latency

    • Real-time responses enabled by edge inference
    • No network round trips
  • Cross-Platform Mobile App

    • Built with React Native
    • Supports iOS and Android

🧩 Technology Stack

  • Frontend

    • React Native
    • TypeScript
  • AI / ML

    • On-device Vision-Language Models (VLMs)
    • RunAnywhere SDK for local inference and model orchestration
  • Platform

    • iOS and Android
    • Native camera and sensor integrations

📱 Use Cases

  • Indoor and outdoor navigation
  • Assistive navigation and accessibility
  • Privacy-sensitive environments
  • Offline or low-connectivity scenarios
  • Edge AI research and experimentation

🔐 Privacy & Security

EdgeVis is designed with privacy as a core principle:

  • No cloud inference
  • No image or sensor data storage
  • No external APIs required for core functionality

🛠️ Development

Prerequisites

  • Node.js
  • React Native development environment
  • RunAnywhere SDK access

Install

npm install

About

On-device AI navigation assistant using RunAnywhere SDK + ONNX VLM runtime integration. Derivative upgrade/testing for Project SightLine (Intel Global AI Award 2025).

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors