Ron Bot Mobile

Technologies & Frameworks

Project Overview

Ron Bot (internal codename Ron) is a consumer hardware product featuring an AI-powered physical desktop companion robot. The product ecosystem communicates directly via localized peer-to-peer protocols and a cloud-assisted realtime broadcast system, completely bypassing intermediary server relays. The physical robot runs a robust, low-latency firmware written in C++ on an ESP32 microcontroller, integrating multiple local sensors, auditory buzzer engines, memory storage, and full WiFi networking capabilities. It interacts with users via a potentiometer-driven 7-page display system (Expressive RoboEyes, Weather Dashboard, Retro Flip-Clock, Pomodoro Timer, Tasks Screen, System Settings, and WiFi QR Pairing) and capacitive touch. The software layer consists of two client companion apps: a native Flutter Windows Desktop Application acting as the cockpit for the bot, and a portrait-locked Flutter Android Mobile Application with a redesigned Material 3 navigation shell. Both apps implement a Smart Connection Router with dual-transport sync (switching between local WebSockets and Supabase Realtime channels) and open-ended chatbot sessions via OpenRouter AI. Additionally, a Next.js 16 storefront provides pre-ordering, waitlisting, and transactional administration, running completely serverless with a Supabase cloud database backend and localized Bangladeshi wallet payment integrations (RupantorPay).

The Challenge & Problem

• Complex Local & Remote Connectivity: Microcontrollers usually require cloud servers for remote access, which introduces latency and operational overhead. Local offline sync is difficult to establish robustly alongside cloud channels.
• Hardened Media Upload Constraints: Microcontroller display hardware (like ESP32 ST7789 TFT screen) has strict LittleFS filesystem limits (~1–2MB) and decoders that cannot support raw high-res or high-fps GIFs/images.
• Screen Jitter from Potentiometers: Analog rotary potentiometers suffer from voltage noise, causing the display pages to rapidly flicker or jitter between selections.
• Payment Idempotency & Redirect Collisions: Bangladeshi local payment gateways (bKash/Nagad/Rocket via RupantorPay APK) suffer from timing delays and callback race conditions between the browser success page redirect and server webhook calls, leading to duplicate orders.

The Engineering Solution

• Smart Connection Router: Designed a dual-transport sync system. The companion app attempts a direct handshake over local WebSocket (ws://ronbot.local/ws) with a 1.5s timeout. If successful, commands and files stream locally with zero lag. If offline locally, it falls back to SSL Cloud Broadcasts using Supabase's Phoenix WebSocket protocol. A background loop queries local availability every 30s to restore peer-to-peer sync.
• Media Processing Pipeline (Hologram Deck): Built a media optimization system on the companion applications. Large image and GIF files are resized dynamically to ≤240×240, frame rates are normalized to a 15fps cap (67ms min delay), and static images are converted to single-frame GIFs. Files are transferred chunk-by-chunk via WebSockets with an automatic retry block (3 attempts with exponential backoff) and base64 encoding to prevent corruption.
• Hysteresis Filter on ESP32: Implemented a software-based hysteresis filter (60 raw units threshold) on the ESP32 firmware. This stabilizes the reading, allowing the user to smoothly switch pages without screen vibration.
• Database-Level Unique Constraint: Enforced database-level unique constraints (rupantor_transaction_id) inside the PostgreSQL orders table. Any simultaneous write is aborted at the database layer, securing payment integrity, while Next.js routes use double confirmation hooks to handle user landing and server webhooks securely.

Key Architecture & Features