The Humanoid Hand Project is a robotic hand that replicates human finger movements in real time by combining computer vision, embedded systems, and mechanical design. The system uses a Raspberry Pi for vision processing, an Arduino for actuation, and a 3D-printed tendon-driven mechanism to physically mimic natural finger curling.

🔹 System Architecture

Raspberry Pi 5 + PiCamera2

Runs a custom MediaPipe hand-tracking pipeline in Python. Finger positions are detected frame-by-frame, translated into binary open/closed states, and mapped to servo angles.

Arduino Uno

Receives servo commands over serial and drives five independent servos, one for each finger (thumb, index, middle, ring, pinky). A tendon + elastic return mechanism provides realistic motion, allowing the hand to curl and extend smoothly.

3D-Printed Hand

Designed with multi-segment joints, tendon routing holes, and elastic anchors to approximate the biomechanics of a human hand. Elastic cords restore the fingers to their neutral (open) position when servos release tension.

🔹 Features

Real-time gesture following: robotic fingers mirror human gestures almost instantly.

Modular codebase: Python scripts for camera testing, servo testing, and full hand control.

Low-cost tendon actuation: elastic return mechanism reduces servo strain and simplifies the design.

Expandable design: architecture can be scaled to a full humanoid arm or integrated into teleoperation systems.

🔹 Applications

Assistive robotics (prosthetics or rehabilitation devices)

Teleoperation and remote manipulation

Humanoid research platforms

Education and STEM outreach

This project demonstrates how vision, control systems, and mechanical prototyping can merge to create accessible humanoid robotics, bridging the gap between human gestures and robotic motion.