Researchers have developed a new framework for decoding high-dimensional finger motion from electromyography (EMG) signals using consumer-grade hardware. This system combines an EMG armband and a webcam to collect a new dataset, EMG-FK, featuring synchronized EMG and 15 finger joint angles from 20 participants. The Temporal Riemannian Regressor (TRR) model, a GRU-based network, processes Riemannian covariance features to achieve state-of-the-art regression accuracy and real-time performance on a Raspberry Pi 5, enabling intuitive control of robotic hands. AI
Summary written by gemini-2.5-flash-lite from 2 sources. How we write summaries →
IMPACT Enables more natural control of prosthetics and AR/XR interfaces through improved EMG decoding.
RANK_REASON Academic paper detailing a new model and dataset for EMG-based motion decoding.