Eyeconic is an advanced AR glasses solution powered by a Raspberry Pi 4 Model B (8GB RAM), offering both offline and online intelligent assistance modes to support real-time translation and smart interactions.
Key Features
Offline Mode
Provides translation capabilities through:
Speech recognition, using lightweight models
OCR-based image translation, enabling users to translate text from captured images
Operates completely without internet connectivity for maximum portability and privacy.
Online Mode
Enables two types of intelligent interactions:
Live Agent Communication: Users can speak directly to a remote assistant for real-time support.
LLM-Powered Chat: Users can interact with a custom LLM via text or voice prompts.
Backend powered by a Django server with REST APIs, interfacing with OpenRouter and Whisper Large v3 for advanced transcription and natural language processing.
Users can:
Record voice prompts or capture images
Send requests to the server and receive intelligent responses
Benefit from fast, reliable AI interaction, even in dynamic environments
Infrastructure & Access Management
One of the major challenges was managing user login and Wi-Fi configuration without traditional input devices (no keyboard or mouse). To solve this:
We built a Flask-based local web portal hosted directly on the Raspberry Pi.
When the device boots, it activates an Access Point (AP) mode.
Users connect to the Pi’s Wi-Fi and access a local webpage to enter:
Wi-Fi credentials
Username and password
This information is securely saved in a JSON file and used by the AR glasses to initialize autonomous operation.
This ensures that Eyeconic becomes a standalone product, requiring no external setup once configured.
Mobile Companion App
We also developed a Flutter-based mobile app that provides:
User registration and authentication
Chat history management
Seamless control of the device experience