RoboBin Hardware Documentation

Overview

RoboBin is an automated waste classification and sorting system built using two ESP32-based microcontrollers:

1. Node32s (ESP32 Board) – Handles sensing (ultrasonic), actuation (servo), display (I2C LCD), and WebSocket communication with the FastAPI server.
2. ESP32-CAM – Captures waste images upon trigger, sends them to the FastAPI server for classification, and receives acknowledgment.

Both devices communicate with the FastAPI backend using WebSockets for real-time synchronization.

---

System Architecture Summary

Component	Role	Communication	Notes
Node32s (ESP32)	Detects object via ultrasonic sensor, controls servo, updates LCD, communicates with server	WebSocket /ws/sensor	Triggers image capture, receives classification
ESP32-CAM	Captures waste image and sends it to the FastAPI server	WebSocket /ws/image	Receives capture signal from Node32s via server
FastAPI Server	Central logic hub that runs ML classification and coordinates both devices	WebSocket + HTTP	Classifies images and returns results

---

Node32s (ESP32) – Sensing & Actuation Unit

Purpose

The Node32s serves as the main control board.
It detects nearby waste using the ultrasonic sensor, triggers classification when necessary, and drives a servo to open/close the bin based on classification.

Functions

• Ultrasonic object detection
• Sends "object_detected" event to server
• Receives classification results ("0" or "1")
• Controls servo motor accordingly
• Displays feedback via I2C LCD
• Supports OTA (Over-the-Air) firmware updates

Pin Connections

Pin	Component	Function
GPIO 4	Ultrasonic Sensor	TRIG
GPIO 5	Ultrasonic Sensor	ECHO
GPIO 13	Servo Motor	PWM Signal
GPIO 21	I2C LCD	SDA
GPIO 22	I2C LCD	SCL

Required Libraries

• WiFi.h
• WebSocketsClient.h
• ESP32Servo.h
• LiquidCrystal_I2C.h
• ArduinoOTA.h (for OTA updates)

---

ESP32-CAM – Image Capture Unit

Purpose

The ESP32-CAM handles image acquisition.
When triggered by the Node32s (via the FastAPI server), it captures an image and sends it to the server for machine learning classification.

Functions

• Waits for "capture" command from the FastAPI server.
• Captures an image using the onboard camera.
• Sends binary image data to /ws/image endpoint.
• Receives classification feedback for confirmation.
• Can also support OTA updates for remote reprogramming.

Pin Notes

Most pins are pre-assigned internally to the OV2640 camera, but ensure:

• GPIO 0 → GND (for flashing only; remove after upload)
• GPIO 4 → LED (optional indicator)
• 5V → VIN
• GND → GND

---

Communication Flow

1. Detection: Node32s detects an object within 20 cm and sends "object_detected" via WebSocket /ws/sensor.
2. Trigger: The FastAPI server forwards "capture" command to the ESP32-CAM via /ws/image.
3. Capture: ESP32-CAM captures and uploads an image to the FastAPI server.
4. Classification: Server runs the ML model and determines the class:
   • "0" → Biodegradable
   • "1" → Non-biodegradable
5. Action: Server sends classification back to Node32s.
6. Servo Response: Node32s moves the servo to the appropriate bin.
7. Resume: Server signals "resume_sensing" to Node32s to restart detection.

---

OTA (Over-The-Air) Updates

Both ESP32 boards support OTA updates using the ArduinoOTA library.

Steps:

1. Connect your PC and the board to the same Wi-Fi network.
2. In Arduino IDE:
   • Go to Tools → Port → Network Ports
   • Select the device (e.g. RoboBin_Node32s or RoboBin_CAM)
3. Click Upload — no USB needed!

Example OTA Configuration

ArduinoOTA.setHostname("RoboBin_Node32s");
ArduinoOTA.setPassword("yourpassword");