Adapting proportional myoelectric-controlled interfaces for prosthetic hands

Area of Science:

• Biomedical Engineering
• Neuroscience
• Robotics

Background:

• Advanced powered hand prostheses with multiple degrees of freedom are transitioning from research to commercial availability.
• Effective control of these multi-actuator prostheses is crucial for users to leverage their full capabilities.
• Current control methods often require significant user adaptation.

Purpose of the Study:

• To investigate the adaptability of human subjects in controlling a dexterous robotic hand using electromyographic (EMG) signals from four hand muscles.
• To determine if learning patterns observed in 2D cursor control tasks are applicable to robotic hand control.
• To assess the efficacy of different muscle-to-actuator mapping strategies.

Main Methods:

• Participants controlled a robotic hand in real-time using EMG signals from four distinct hand muscles.
• Various muscle-to-actuator mapping strategies were employed during the study.
• Learning progress was quantified and compared to established 2D cursor control learning metrics.

Main Results:

• Human subjects demonstrated a significant ability to learn real-time control of the dexterous robotic hand.
• The learning process for robotic hand control mirrored that of 2D cursor control, even with non-intuitive muscle mappings.
• Adaptation occurred across different mapping strategies, indicating user flexibility.

Conclusions:

• Electromyographic (EMG) control is a viable and adaptable method for operating advanced prosthetic hands.
• Users can rapidly learn to intuitively control complex robotic prostheses, similar to cursor control.
• This finding supports the development of more user-friendly and functional prosthetic devices.