Trust sensor interface for improving reliability of EMG-based user intent recognition

Area of Science:

• Biomedical Engineering
• Rehabilitation Engineering
• Signal Processing

Background:

• Electromyographic (EMG) signals are crucial for decoding user intent in prosthetic devices.
• Signal contamination by disturbances like motion artifacts and baseline shifts compromises prosthetic control accuracy and safety.

Purpose of the Study:

• To develop a novel trust sensor interface (TSI) for detecting disturbances in EMG signals.
• To enhance the reliability of user intention recognition (UIR) algorithms in prosthetic control systems.

Main Methods:

• Proposed a TSI with two modules: an abnormality detector and a trust evaluation component.
• Implemented a system where the UIR algorithm dynamically adjusts based on TSI output.
• Conducted experiments on an able-bodied subject to validate the TSI's performance.

Main Results:

• The TSI accurately detected two common disturbances: motion artifacts and baseline shifts with high accuracy and low latency.
• Demonstrated significant improvement in the reliability of the user intention recognition (UIR) algorithm.
• Validated the TSI's effectiveness in real-world application scenarios.

Conclusions:

• The proposed trust sensor interface (TSI) effectively addresses EMG signal disturbances.
• The TSI enhances the safety and reliability of EMG-based prosthetic control systems.
• This approach offers a promising solution for robust human-machine interfaces in prosthetics.