Project / Case study

Robotic Arm with Computer Vision

This project uses a webcam and hand tracking to control a robotic arm. MediaPipe detects hand landmarks, Python interprets gestures, and commands travel over serial to an Arduino-driven servo controller.

At a glance

complete

  • Python
  • OpenCV
  • MediaPipe
  • Arduino
  • PCA9685
  • Serial communication
View the repository ↗
A robotic arm computer vision project being demonstrated at an exhibition

01 / Research question

What it asks

How can natural hand gestures become reliable controls for a physical robotic arm?

02 / Experimental design

Scope

A webcam captures a single hand, MediaPipe extracts its landmarks, and the Python application identifies finger counts, fists, thumb-to-index distance, and left- or right-hand context. The resulting command is sent to an Arduino over USB serial.

03 / Evaluation

How it measures

The Arduino sketch updates the arm servos through an Adafruit PCA9685 PWM Servo Driver. Gesture mappings provide separate movement commands for left and right hands, including stop, directional motion, and gripper-related actions.