EMG-Based Gesture Recognition
Overview
The field uses surface electromyography (EMG)—electrical activity measured at the skin—to infer gestures and movement intent, enabling hands-free control of prostheses, assistive devices, and interactive systems. Approaches range from time- and frequency-domain features with classifiers such as LDA and SVM to CNNs, recurrent and graph networks, and multimodal models that fuse EMG with inertial, pressure, or visual signals. Current work explores self-supervised pretraining, cross-user and cross-session adaptation, and compact models that can run with low latency on wearable hardware. A central challenge is maintaining reliable performance across people, sensor placements, sessions, and spontaneous movements without requiring extensive labeled calibration data.
Latest papers
26 papers3 papers in the last four weeks, down 40% on the four weeks before. 0.0% of all new papers.