Robotics · Computer Vision · Human–Machine Interaction
Project MIMIC
A robotic hand that moves with you.
MIMIC translates human hand movements into real-time robotic motion. Using computer vision to track finger positions, the system estimates individual finger bends and sends those values to an Arduino-controlled robotic hand — a direct connection between human movement and physical machinery, without gloves, controllers, or wearable sensors.
Overview
A camera watches your hand. Computer vision locates the joints of each finger, and from those positions MIMIC estimates how far each finger is bent. Those bend values drive a physical robotic hand so that it follows along with your own hand in real time.
Technical approach
- Hand landmarks are detected from a live camera feed with computer-vision hand tracking.
- For each finger, the angles between its joints are measured and mapped to a bend percentage from fully open to fully closed.
- The per-finger bend values are sent to an Arduino-controlled robotic hand, which actuates to match.
- No gloves, controllers, or wearable sensors are required — the camera is the only input.
The live demo below runs the same tracking and finger-bend estimation entirely in your browser. The camera feed never leaves your device.
Live demo
Allow camera access, then show your hand to the camera. All processing is local.
Your camera feed never leaves your device — all processing runs locally in the browser.
Media
Build photos, bench footage, and interactive animations will be added here.
Reserved for project photos, video, and a Rive animation.