Automated Grasp Recognition Using sEMG: Recent Advances, Challenges, and Future Developments

• Published in: IEEE transactions on instrumentation and measurement Vol. 74; pp. 1 - 17
• Main Authors: Sharma, Shivam, Newaj Faisal, Kazi, Raj Sharma, Rishi
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.01.2025; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Artificial neural networks; Automated grasp recognition (AGR); Automatic control; Automation; deep neural networks (DNNs); Degrees of freedom; Electromyography; End effectors; Estimation; Feature extraction; Focusing; Force; Gesture recognition; Grasping (robotics); Industries; Machine learning; machine learning (ML); Man-machine interfaces; model-based approach; Multiresolution analysis; Prostheses; Reviews; Robot sensing systems; Sensors; State-of-the-art reviews; surface electromyography (sEMG); Virtual reality
• DOI: 10.1109/TIM.2024.3497179
• ISSN: 0018-9456

Surface electromyography (sEMG)-based automated grasp recognition (AGR) has emerged as a vital technology in the field of automatic control, human-machine interfaces, prosthetics, virtual reality (VR), etc. Grasp recognition comes under the category of hand gesture recognition (HGR). However, due to its unique characteristics, traits, and advanced applications, such as a robotic hand with 3-D-grasping capability, gaming, etc., it differs from other hand gestures in the case of force estimation and degree of freedom (DOF). This article provides a comprehensive review of available state-of-the-art methodologies for sEMG-based AGR. The review covers sensing modalities, datasets focusing on different grasps types (including power, precision, cylindrical, spherical, tripod, lateral, hook, and palmar grasps), sEMG acquisition systems, pre-processing techniques, multiresolution analysis (MRA), feature extraction process, and identification systems focusing on machine learning (ML), deep neural networks (DNNs), and model-based approaches. This review provides a detailed year-by-year chronological analysis and comparison of grasp recognition techniques with specific focus on number of subjects and types of grasps. Furthermore, some open research issues have been pointed out from the reviewed literature, and possible future prospects for these challenges have also been presented. Finally, several industry domains that can incorporate sEMG-based AGR systems in future are also discussed.