An EMG-Based Deep Learning Approach for Multi-DOF Wrist Movement Decoding

• Published in: IEEE transactions on industrial electronics (1982) Vol. 69; no. 7; pp. 7099 - 7108
• Main Authors: Yang, Dapeng, Liu, Hong
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.07.2022; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Adaptation models; Artificial intelligence; Artificial neural networks; Convolutional neural network (CNN); Decoding; Deep learning; deep learning (DL); Degrees of freedom; Electromyography; End effectors; Feature extraction; Hand (anatomy); Human motion; Model accuracy; motion decoding; myoelectric signal; Myoelectricity; Prostheses; Rehabilitation; Robot arms; Robotics; Robots; Three-dimensional displays; Training; Wrist
• ISSN: 0278-0046; 1557-9948
• DOI: 10.1109/TIE.2021.3097666

In robotics, decoding complex human movements of multiple degrees of freedom from surface electromyography (sEMG) remains challenging. Recently, the rapid development of artificial intelligence (AI) technology provides a new solution to this problem. In this article, we propose an AI-based framework that consists of a series of deep learning approaches for achieving a precise and robust decoding on three-dimensional (3-D) wrist movements. A previously developed device (wrist movement detector) was utilized to tag the myoelectric signals with 3-D kinematic labels. A public dataset (HIT-SimCo) was established, wherein the sEMG signals were collected from diverse wrist movements and multiple subjects. A lightweight convolutional neural network was constructed, which can extract the motion-related features directly from raw sEMG, and make motion predictions in an end-to-end manner. In addition, several data augmentation strategies were explored to improve the robustness of the model against environmental variations; and a fine-tuning policy was proposed to further improve the subject-specific accuracy. The decoding model was finally tested in three scenarios: a robotic arm/hand system performing daily-living tasks (pouring, screwing, etc.), a supernumerary robotic hand playing blocks-building, and a virtual prosthesis conducting motor rehabilitation training.