A Soft Exoskeleton Glove for Hand Bilateral Training via Surface EMG

Area of Science:

• Biomedical Engineering
• Rehabilitation Robotics
• Human-Computer Interaction

Background:

• Traditional rigid exoskeletons present comfort and pressure issues, potentially hindering natural hand motion.
• There is a need for comfortable, low-cost assistive devices for hand rehabilitation and bilateral training.

Purpose of the Study:

• To develop and evaluate a low-cost soft exoskeleton glove (SExoG) system.
• To enable bilateral training by mirroring hand motions of the non-paretic hand using surface electromyography (sEMG) signals.

Main Methods:

• Customization of soft actuator geometrical parameters and redesign of their structure.
• Determination of air-pump pressure values for four distinct hand motions (extension, rest, spherical grip, fist).
• Development of a two-step hybrid model (neural network + state exclusion algorithm) for sEMG-based hand motion recognition.

Main Results:

• The SExoG system successfully supported four hand motions with specific pressure values (-2, 0, 40, 70 KPa).
• The hybrid model achieved a high mean accuracy of 98.7% in recognizing hand motions from sEMG signals.
• Experimental validation with four subjects demonstrated the system's effectiveness.

Conclusions:

• The proposed SExoG system offers a promising, low-cost solution for hand rehabilitation.
• The system effectively mirrors non-paretic hand motions, facilitating bilateral training.
• The developed sEMG recognition model demonstrates high accuracy and reliability.