An EMG-Driven Musculoskeletal Model for Estimating Continuous Wrist Motion.

Area of Science:

• Biomedical Engineering
• Robotics
• Neuroscience

Background:

• Electromyography (EMG) based continuous wrist joint motion estimation shows promise for assistive robots.
• Conventional methods lack interpretability of the neuro-muscular-skeletal relationship.

Purpose of the Study:

• To propose and validate an EMG-driven musculoskeletal model for accurate continuous wrist joint motion estimation.
• To interpret neuro-commands and their relation to joint motion.

Main Methods:

• Developed an EMG-driven musculoskeletal model integrating muscle-tendon and joint kinematic models.
• Utilized a genetic algorithm for optimizing subject-specific physiological parameters.
• Trained and validated the model using EMG and motion capture data.

Main Results:

• Achieved mean root-mean-square errors below 18° across various motion trials (single flexion/extension, continuous cycle, random).
• Obtained a mean coefficient of determination over 0.9, indicating high prediction accuracy.
• Demonstrated accurate tracking performance aligned with user's intended movements.

Conclusions:

• The proposed EMG-driven musculoskeletal model accurately estimates continuous wrist joint motion.
• This approach enhances the interpretability of EMG signals for controlling assistive devices.
• The model offers a promising advancement for intention-driven robotic applications.