• 

  Chatbot AI

• 

  System Architecture Overview

• 

  Electronic Components Overview

• 

  3D Prototype Hardware Design

• 

  3D Prototype Hardware Design

• 

  System Architecture Overiew

• 

  Layered System Architecture

• 

  Dashboard Website

• 

  Dashboard Website

Inspiration The idea came from wanting to create a comprehensive IoT access control system that goes beyond simple door locks. I wanted to integrate multiple safety features: automated obstacle detection for preventing accidents, environmental monitoring for fire/humidity alerts, IR remote and emergency button access, AI-powered chatbot assistance, and seamless remote management with web dashboard—all on a single microcontroller platform.

What it does The ESP32 Smart Door Control System is a multi-functional access control solution featuring:

• IR remote control for door unlock (OK button), LED toggle (* button), and fan toggle (# button)
• Emergency exit button that opens the door for a configurable duration (default 10 seconds)
• Automated servo lock mechanism (configurable angles: default 0° closed, 90° open)
• Ultrasonic obstacle detection that auto-activates lighting when someone approaches (threshold configurable, default 10 cm)
• Real-time temperature & humidity monitoring with configurable alert thresholds (default 30°C temp, 70% humidity; auto-opens door at 85°C or 85%)
• WiFi-enabled web dashboard with real-time WebSocket updates, controls, settings, AI chat, about, and contact tabs
• AI chatbot assistant using Google Gemini for natural language queries and commands (e.g., /open, /led on)
• Voice input support for chatbot via browser speech recognition
• Visual feedback via 20x4 I2C LCD display showing system status, sensor data, alerts, countdown timer, LED/fan status
• Audio alerts through active buzzer for access events, door open, and temp/humidity alerts with different beep patterns
• Relay-controlled lighting (auto-on with obstacle)
• Status LED with blinking patterns (rapid flash when door open)
• Toggle LED and PWM-controlled DC fan with physical buttons and remote/web control
• Data logging (up to 100 entries) with CSV export via web
• Email alerts via Formspree for temp/humidity thresholds (with 5-minute cooldown)

How we built it Hardware Integration:

• ESP32-WROOM-32 as the central controller (dual-core Tensilica LX6 @ 240MHz)
• HC-SR04 ultrasonic sensor for distance detection (trig GPIO19, echo GPIO18)
• DHT22 sensor for precise temperature (±0.5°C) and humidity (±2% RH) measurements (data GPIO25 with 4.7kΩ-10kΩ pull-up)
• SG90 servo motor controlled via PWM for lock mechanism (GPIO26)
• 5V relay module for automated lighting control (GPIO27)
• IR receiver for remote control (GPIO14)
• Emergency exit button (GPIO34, active LOW with pull-up)
• Active buzzer for alerts (GPIO15)
• Status LED (GPIO5)
• 20x4 I2C LCD for real-time status display (address 0x27, SDA GPIO21, SCL GPIO22)
• Toggle LED (GPIO13) with button (GPIO35, pull-up)
• DC fan motor with PWM (GPIO32, channel 8, 25kHz freq)
• Fan button (GPIO23, pull-up)

Software Architecture:

• Developed in Arduino C++ using ESP32 core libraries
• AsyncWebServer and AsyncTCP for non-blocking web interface with WebSocket (/ws)
• IRremote library for IR decoding (updated to new version)
• DHT library for sensor reading with rate limiting (every 2 seconds)
• ESP32Servo for servo control
• LiquidCrystal_I2C (ESP32-compatible) for LCD
• Preferences for persistent storage of configurable settings (thresholds, angles, timing, API keys)
• HTTPClient for Formspree email alerts
• Multi-threading utilizing FreeRTOS tasks on both ESP32 cores
• State machine for door control (manual, auto-ventilation modes)
• Debouncing for buttons (50ms delay)
• Data logging in circular buffer with CSV generation
• AI integration via Gemini API in web client (generateContent endpoint)
• Voice recognition using Web Speech API in browser
• WiFi AP mode (SSID: ESP32SEG, password: letmeinplease)

Key Technical Implementations:

• Custom handling for DHT22 with timestamp-based interval checks
• Proper pull-up configurations (e.g., 4.7kΩ on DHT data, external 10kΩ on buttons)
• Power budgeting to avoid brownouts (separate power for actuators)
• Ultrasonic pulse timing with timeout (30ms) and distance calculation (temperature-compensated implicitly via stable readings)
• Responsive HTML/CSS/JS web dashboard with tabs, real-time updates via WebSocket, AJAX, toast notifications
• Command parsing in chatbot for system control (e.g., regex matching for /open, /fan on)
• Different buzzer patterns: fast for door open, slow for alerts, rapid for multiple, continuous for manual
• LCD optimization: update only changed lines, periodic backlight refresh
• Settings persistence and API key management (Gemini, Formspree)

Challenges we ran into

1. Voltage level compatibility: HC-SR04 outputs 5V ECHO signal, but ESP32 GPIO is only 3.3V tolerant. Solved with precision voltage divider (1kΩ + 2kΩ resistors).

2. Power consumption: Initial design tried powering servo and relay from ESP32's 3.3V rail, causing brownouts. Redesigned power distribution using VIN (5V USB) rail with proper current budgeting.

3. DHT22 timing sensitivity: Sensor requires minimum 2-second intervals between reads. Implemented timestamp-based rate limiting to prevent corrupt readings.

4. Ultrasonic noise: Raw HC-SR04 readings showed spurious spikes. Implemented timeout and default to 999cm on failure (median filter from original retained conceptually).

5. Web server blocking: Initial synchronous web server caused main loop delays. Switched to AsyncWebServer for true non-blocking operation.

6. Memory constraints: ESP32's 520KB SRAM filled quickly with large web pages and data log. Optimized HTML/CSS, used PROGMEM for static strings, implemented chunked HTTP responses, and limited log to 100 entries.

7. IR library compatibility: Updated to new IRremote.hpp for better decoding and repeat flag handling.

8. PWM channel conflicts: Servo and fan PWM conflicted; reassigned fan to channel 8.

9. Browser voice recognition: Handled browser support checks and error states for Web Speech API.

10. Alert spam prevention: Added 5-minute cooldown for email alerts and edge detection for state changes.

Accomplishments that we're proud of

• Successfully integrated over 10 different components (sensors, actuators, displays, buttons, remote) on a single microcontroller
• Achieved robust door control with multiple modes: manual (button/IR/web/AI), auto-ventilation with no timer
• Designed a professional-grade web dashboard with tabbed interface, real-time updates, voice-enabled AI chatbot, and CSV export
• Implemented AI assistance with natural language and command parsing, integrated with system status
• Created configurable system with web-based settings for thresholds, angles, timing, and API keys
• Added voice control and email notifications for enhanced usability
• Implemented data logging and export for analysis
• Maintained modular code architecture for easy feature additions, with comprehensive debugging via Serial
• Optimized for stability: debouncing, rate limiting, I2C traffic reduction

What we learned

• Deep understanding of ESP32 architecture: dual-core processing, FreeRTOS, peripheral buses (I2C, PWM channels), power distribution, pin constraints (e.g., input-only pins, strapping pins)
• Hands-on experience with multiple communication protocols: I2C (LCD), 1-Wire (DHT22), PWM (servo/fan), GPIO (matrix scanning, interrupts implicitly via polling)
• Importance of electrical safety: voltage level translation, current limiting, proper grounding, pull-ups
• Web development for embedded systems: asynchronous JavaScript, WebSocket, AJAX polling, responsive design, client-side API calls (Gemini, speech)
• Signal processing: debouncing, edge detection, pulse timing, noise handling
• Power management and current budgeting for battery-powered applications
• API integration: Google Gemini for AI, Formspree for emails, persistent storage with Preferences
• Browser features: Web Speech API for voice, toast notifications, dynamic UI updates
• System design: State tracking (e.g., auto vs manual open), alert patterns, data logging structures

Built With

• 1-wire
• 1-wire-(for-dht22-sensor)
• 1k?
• 2k?
• 4x4-matrix-keypad
• 5v-relay-module
• 802.11
• active-buzzer
• ap
• arduino
• asynctcp
• asyncwebserver
• breadboad
• c++
• core
• cp210x
• css3
• dht22-sensor
• electronic
• embedded
• esp32
• esp32-wroom-32
• espasyncwebserver-(for-non-blocking-web-server)
• espressif
• firebase
• formspree
• formspree-(for-contact-form-integration)
• freertos
• gemini
• google
• google-gemini-(for-ai-chatbot-integration)-databases:-preferences-(for-storing-system-configurations-locally-on-the-esp32)-apis:-google-gemini-api-(for-ai-powered-chatbot)
• gpio
• hc-sr04-ultrasonic-sensor
• html5
• httpclient
• httpclient-(for-sending-email-alerts)-other-technologies:-i2c-(for-lcd-communication)
• i2c
• i2c-lcd-(20x4)
• ieee
• if-applicable)
• irremote.hpp-(for-ir-remote-control)-platforms:-esp32-wroom-32-(microcontroller-platform)-cloud-services:-firebase-(for-hosting-and-database
• javascript
• javascript-(for-web-interface)-frameworks:-freertos-(for-multi-core-task-handling)
• lcd
• libraries
• microcontroller
• progmem
• pwm
• pwm-(for-servo-motor-control)
• resistor
• sg90-servo-motor
• usb
• websocket
• wi-fi
• wifi
• wire