EMG-driven shared human-robot compliant control for in-hand object manipulation in hand prostheses.

Area of Science:

• Robotics
• Biomedical Engineering
• Neuroscience

Background:

• Limited dexterity of current prosthetic hands hinders widespread adoption by amputees.
• Existing prostheses often lack the fine manipulation capabilities needed for complex tasks.
• Electromyographic (EMG) decoding offers a pathway to intuitive prosthetic control.

Purpose of the Study:

• To investigate a novel shared control strategy for robotic prosthetic hands.
• To combine autonomous force control with EMG decoding for robust in-hand object manipulation.
• To improve the dexterity and utility of prosthetic hands for individuals with trans-radial amputations.

Main Methods:

• A three-day longitudinal study with eight healthy subjects controlling a 16-degrees-of-freedom robotic hand.
• EMG decoding from forearm muscles for proportional and simultaneous finger control.
• Shoulder EMG electrodes for inferring desired object rotation, combined with an autonomous controller for stabilization and rotation.

Main Results:

• Robotic assistance reduced task failures by 41%.
• Shared control with incremental shoulder EMG decoding led to faster task completion (median = 16.9s).
• Subjects achieved 50% fewer failures after just one training session.

Conclusions:

• Shared control strategies offer an alternative to pure EMG decoding for prosthetic hand control.
• Delegating force control to the prosthesis improves reaction time and precision.
• This approach can enhance prosthetic hand dexterity, enabling fine insertion tasks and restoring arm functionality.