Happy Hands: Vision-Guided Finger Rehabilitation Sleeve
Team: Maysarah Sukkar, Yousif Alhajji, Benjamin Pedi, Lily Snow, Kaedin Kurtz
Course: ME571 Medical Robotics, Boston University
Overview
Stroke and neurological patients rebuild hand function through repeated, guided movement. Mirror-therapy gloves already guide the impaired hand using the healthy one, but they require a second glove, which adds bulk and makes fitting harder. Happy Hands replaces the second glove with a webcam, so the healthy hand guides the impaired one without wearing anything.
System Design
Sensing: ml5.js pose estimation tracks finger joint keypoints on the healthy hand from a standard webcam.
Processing: MATLAB converts the keypoints to joint angles and decides whether each finger should flex, extend, or hold.
Actuation: An Arduino Uno drives two stepper motors on a wrist cuff. Each motor winds a string running through a breathable elastic finger sleeve, pulling the finger into flexion.
Extension and resistance: Springs on the back of the hand assist extension and resist flexion. Most rehab devices ignore extension. The springs are swappable, so resistance can scale with recovery.
Feedback: A live plot shows users their own joint angles over time.
Results
In a five-person NASA-TLX evaluation, tracking and closed-loop control performed reliably with low latency. The main issues were comfort: the wrist-mounted motors were heavy, and the springs had to be attached by tying knots. Below are the detailed paper and video on our system and research.
Comments