Hand Tracking Servo

Real-Time Hand Tracking Servo Control

This project demonstrates real-time servo control using computer vision and hand tracking. A webcam detects hand movements using MediaPipe, and the data is sent to an Arduino Nano, which controls multiple servo motors based on finger positions.

Main Features:

Required Components:

Wiring:

How it works:

The Python script uses MediaPipe to detect hand landmarks from the webcam. It determines which fingers are up or down and converts this into a binary string. This string is sent via serial communication to the Arduino. The Arduino reads the data and moves the corresponding servos.

Code for Arduino


#include <Servo.h>

Servo s1, s2, s3, s4, s5;

void setup() {
  Serial.begin(9600);

  s1.attach(3);
  s2.attach(5);
  s3.attach(6);
  s4.attach(9);
  s5.attach(10);
}

void loop() {
  if (Serial.available() >= 5) {

    char f1 = Serial.read();
    char f2 = Serial.read();
    char f3 = Serial.read();
    char f4 = Serial.read();
    char f5 = Serial.read();

    s1.write(f1 == '1' ? 180 : 0);
    s2.write(f2 == '1' ? 180 : 0);
    s3.write(f3 == '1' ? 180 : 0);
    s4.write(f4 == '1' ? 180 : 0);
    s5.write(f5 == '1' ? 180 : 0);
  }
}

Code for Python


import cv2
import mediapipe as mp
import serial
import time

arduino = serial.Serial('COM4', 9600)
time.sleep(2)

mp_hands = mp.solutions.hands
hands = mp_hands.Hands(
    min_detection_confidence=0.7,
    min_tracking_confidence=0.7
)

mp_drawing = mp.solutions.drawing_utils
cap = cv2.VideoCapture(0)

def detect_fingers(lm):
    fingers = []

    fingers.append(1 if lm.landmark[4].x < lm.landmark[3].x else 0)
    fingers.append(1 if lm.landmark[8].y < lm.landmark[6].y else 0)
    fingers.append(1 if lm.landmark[12].y < lm.landmark[10].y else 0)
    fingers.append(1 if lm.landmark[16].y < lm.landmark[14].y else 0)
    fingers.append(1 if lm.landmark[20].y < lm.landmark[18].y else 0)

    return fingers

while True:
    success, img = cap.read()
    if not success:
        break

    img = cv2.flip(img, 1)
    rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
    results = hands.process(rgb)
    img = cv2.cvtColor(rgb, cv2.COLOR_RGB2BGR)

    if results.multi_hand_landmarks:
        for hand_landmarks in results.multi_hand_landmarks:

            mp_drawing.draw_landmarks(
                img,
                hand_landmarks,
                mp_hands.HAND_CONNECTIONS
            )

            fingers = detect_fingers(hand_landmarks)
            data = ''.join(map(str, fingers))
            arduino.write(data.encode())

    cv2.imshow("Hand Tracking", img)

    if cv2.waitKey(1) & 0xFF == 27:
        break

cap.release()
arduino.close()
cv2.destroyAllWindows()