Context-dependent adaptation improves robustness of myoelectric control for upper-limb prostheses

Area of Science:

• Biomedical Engineering
• Robotics
• Neuroscience

Background:

• Dexterous upper-limb prostheses aim to restore grasping function.
• Current myoelectric control lacks reliable feed-forward mechanisms.
• Improving control robustness and reliability is crucial for prosthesis users.

Purpose of the Study:

• To enhance the robustness and reliability of myoelectric control for upper-limb prostheses.
• To develop and evaluate a context-driven myoelectric control (cxMYO) scheme.
• To integrate proprioception and exteroception sensor data for improved control.

Main Methods:

• Developed a context-driven myoelectric control (cxMYO) scheme.
• Incorporated inertial measurement unit (proprioception) and force/grip aperture (exteroception) sensors.
• Performed online evaluation in able-bodied subjects comparing cxMYO with traditional machine-learning myoelectric control (MYO) across three functional tasks.

Main Results:

• cxMYO significantly decreased unwanted commands compared to MYO.
• The number of dropped objects per round was zero with cxMYO across all tasks.
• Successful transfer rates increased significantly in two out of three functional tasks.
• Subjects reported an improved user experience.

Conclusions:

• Integrating on-board sensor information improves myoelectric control robustness.
• The cxMYO scheme offers a simple, non-invasive, and cost-effective approach.
• This is the first online evaluation of such an integrated sensor-modulation system for myoelectric control.