Project 03 / 2026

Smart Glass

IoT & Backend Developer/ ESP32/ C++/ TypeScript/ WebSocket/ IoT

Year

2026

Role

IoT & Backend Developer

Technologies

ESP32, C++, TypeScript, WebSocket, IoT

About this project

Smart Glass combines embedded display firmware and a Node.js backend to support low-latency text streaming for wearable translation scenarios. The IoT side handles queueing, buffer management, and OLED scrolling; the server owns HTTP state and WebSocket delivery.

Overview

Smart Glass focuses on realtime text visualization for wearable or near-eye display use cases. The embedded app receives incoming text chunks, queues them safely, renders readable text on an OLED display, and scrolls long messages smoothly. The backend provides REST endpoints and a WebSocket channel for live translation and navigation data.

My Role

Built the embedded display service, queue-driven rendering loop, OLED scrolling behavior, and backend WebSocket lifecycle. The work required balancing readable UX on constrained hardware with reliable low-latency server communication.

Key Features

06 Items

Feature

Realtime Text Ingestion

Incoming text is accepted through a push API, buffered through a queue, and consumed on a controlled display loop.

Feature

OLED Scroll Renderer

Long messages are measured in pixels and rendered as a seamless horizontal scrolling loop with configurable gap spacing.

Feature

Bounded Display Buffer

Sliding-window trimming keeps the display buffer within memory limits while preserving recent readable content.

Feature

ESP32 PlatformIO Build

Firmware targets a Freenove ESP32-S3 WROOM board with Arduino, Adafruit GFX, SSD1306, ArduinoJson, and ArduinoWebsockets.

Feature

WebSocket Translation Channel

Express server attaches a WebSocket server at /ws and initializes translation handling for each connection.

Feature

Graceful Server Shutdown

Shutdown handler closes active WebSocket clients, the WebSocket server, and the HTTP server to avoid orphaned connections.

Tech Stack

Firmware

C++PlatformIOArduinoESP32-S3Adafruit GFXAdafruit SSD1306

Backend

Node.jsTypeScriptBunExpresswsSocket.IOdotenv

Realtime

WebSocketPicovoice CheetahGoogle Cloud SpeechAxiosMulter

The Challenge

Wearable displays have tight memory, timing, and readability constraints. Incoming translation text can arrive faster than the display should update, and long sentences must remain readable on a small OLED without flicker or heap pressure.

The Solution

Decoupled producers from rendering with a queue, used fixed pop and render intervals, kept the text buffer bounded, and measured pixel width before enabling scrolling. On the server, WebSocket connection handling is isolated from REST state and includes explicit cleanup paths.

Results

04 Metrics

Impact