Predicting Walking Intentions using sEMG and Mechanical sensors for various environment.

Area of Science:

• Robotics
• Biomechanics
• Biomedical Engineering

Background:

• Exoskeleton robot control is challenged by predicting user motion intentions.
• Mechanical sensors detect motion during movement, while electromyographic (EMG) signals indicate muscle activation before movement.
• Utilizing EMG signals can anticipate motion, reducing control delays and user resistance.

Purpose of the Study:

• To enhance exoskeleton robot control by predicting user motion intentions.
• To investigate the efficacy of combining surface electromyographic (sEMG) and mechanical sensors for gait cycle identification.
• To analyze sensor combination variations and leg information (one vs. two legs) across different gait phases.

Main Methods:

• Employed surface electromyographic (sEMG) signals alongside mechanical sensors (kinetic and kinematic).
• Classified sensor combinations and analyzed data from one and two legs.
• Examined gait periods, specifically before heel contact and toe-off.
• Evaluated prediction accuracy based on time intervals (100ms and 200ms) before heel contact.

Main Results:

• Achieved high classification accuracy (96.8% for one leg, 98.6% for two legs) using sEMG, kinetic, and kinematic sensors before heel contact.
• Demonstrated successful gait prediction by dividing time intervals before gait initiation.
• Obtained average accuracies of 84.4% (100ms interval) and 90.9% (200ms interval) before heel contact.

Conclusions:

• Combining sEMG and mechanical sensors provides accurate prediction of user motion intentions for exoskeleton control.
• The proposed method effectively identifies gait phases and predicts movement, enhancing real-time control.
• This approach significantly reduces control delays and improves the user experience with exoskeleton robots.