🎯 PID-Based Object Balancing System 🎯

📌 Introduction

This system utilizes PID (Proportional-Integral-Derivative) control to balance an object on a dynamic platform. A smartphone camera tracks the object’s position in real time and sends data to a microcontroller via UART for servo motor adjustments.

📷 Camera → Image Processing → PID Error Calculation → Servo Control

🚀 Key Features

✅ Real-time object tracking using OpenCV.
✅ Flexible PID control for quick system stabilization.
✅ UART connection with a microcontroller for instant response.
✅ Manual PID Tuning – fine-tune each parameter for optimal performance.
🔹 Note: The machine learning model is only used to verify whether the object is balanced. The system's control is entirely handled by the PID algorithm.

---

🛠️ System Requirements

🔌 Hardware

• Windows 10 PC (as the software is primarily developed for Windows 10).
• Smartphone camera for object tracking (wired connection preferred to reduce latency).
• Microcontroller (ESP32, Arduino, STM32, etc.) connected via UART.
• Servo motors for platform adjustment.

💻 Software

Install the required Python libraries:

pip install opencv-python numpy matplotlib joblib pyserial

---

🎮 Usage Instructions

1️⃣ Connect the smartphone camera to the PC

Use IVCam (Android) or DroidCam for video streaming.

2️⃣ Load the trained model

from joblib import load
model = load("random_forest_model.joblib")

3️⃣ Open UART communication with the microcontroller

import serial
ser = serial.Serial('COM7', baudrate=115200, timeout=0)

4️⃣ Run the main program

python PID.py

5️⃣ Adjust PID parameters using shortcut keys

Key	Function
W/A/S/D	Move tracking window
1-4	Change setpoint
F	Toggle tracking area
Q	Exit program

---

🎛️ Manual PID Tuning

Fine-tune PID parameters step by step:

🔹 Step 1: Adjust Kp – increase until the system responds quickly without excessive oscillations.
🔹 Step 2: Adjust Kd – add to reduce oscillations and stabilize movements.
🔹 Step 3: Adjust Ki – add only if there is a steady-state error but avoid high values to prevent lag.

🔥 Tuning Tip: If excessive oscillations occur, disable one term (Kd or Ki) and retune from scratch!

Example PID parameters:

PID_params = {
    'x': {'kp': 2.5, 'ki': 0.2, 'kd': 45},
    'y': {'kp': 2.5, 'ki': 0.2, 'kd': 45}
}

---

🔧 ESP32 Firmware

🔹 Overview

The ESP32 firmware performs the following tasks:

• Receives object position data from the PC via UART.
• Calculates error and updates servo angles based on PID parameters.
• Controls two servo motors to maintain balance.

🔹 Workflow

1. Initialize ESP32, assign servo pins, and set up Serial communication.
2. Receive object position data from the PC via Serial.
3. If no object is detected, set error values to -1.
4. If valid data is received, update servo angles accordingly.
5. Repeat the loop.

🔹 UART Communication Format

ESP32 receives data from the PC in the format:

x_valuex y_value

• x_value and y_value represent the object's coordinates.
• If the object is not detected, ESP32 receives "no".

🔹 ESP32 Code

#include <Servo.h>
#include <Wire.h>

static const int servoPin1 = 16;
static const int servoPin2 = 17;
Servo servo1, servo2;

float x_error = -1, y_error = -1;
unsigned long previousMillis = 0;

void setup() {
  Serial.begin(115200);
  servo1.attach(servoPin1);
  servo2.attach(servoPin2);
  servo1.write(55);
  servo2.write(55);
}

void loop() {
  unsigned long currentMillis = millis();
  if (currentMillis - previousMillis >= 1) {
    previousMillis = currentMillis;
    processData();
    servo1.write(x_error);
    processData();
    servo2.write(y_error);
  }
}

void processData() {
  if (Serial.available() > 0) {
    String input = Serial.readStringUntil('\n');
    if (input == "no") {
      x_error = -1;
      y_error = -1;
    } else {
      int separatorIndex = input.indexOf('x');
      if (separatorIndex != -1) {
        x_error = input.substring(0, separatorIndex).toInt();
        y_error = input.substring(separatorIndex + 1).toInt();
      }
    }
  }
}

🔹 Future Enhancements

• Implement signal filtering to reduce noise from the camera.
• Optimize the PID algorithm for faster response times.
• Improve Serial communication for efficient data handling.

---

📸 Images & Videos

📌 System Images:

📌 System Demo Videos:

• Adjust balance position via keyboard
• Impressive balancing performance
• Upcoming update – Setpoint adjustment via mouse clicks

---

📜 License 🚀 Open-source project – Free to use for educational & commercial purposes

Created by Trần Trọng Phúc 📧 Contact: trantrongphucttp27@gmail.com

---

🔥 If you find this project useful, don't forget to ⭐ Star the repo!