Shared Prosthetic Control Based on Multiple Movement Intent Decoders

Area of Science:

• Biomedical Engineering
• Neuroprosthetics
• Human-Computer Interaction

Background:

• Multiple movement intent decoders exist, each with unique strengths and weaknesses.
• Combining decoders may enhance prosthetic limb control performance beyond individual methods.

Purpose of the Study:

• To present and evaluate a shared controller framework for prosthetic limbs using multiple volitional movement intent decoders.
• To develop and validate an algorithm for combining decoder estimates to control prosthetic devices.

Main Methods:

• A shared controller framework was developed, integrating a Kalman filter decoder and a multilayer perceptron classifier decoder.
• Online experiments were conducted with amputee and intact-arm subjects controlling a virtual limb in real-time.

Main Results:

• The shared controller demonstrated statistically significant improvements compared to individual decoders.
• Key performance metrics showed increased time-in-target and decreased unintended movements.

Conclusions:

• The shared controller effectively combines the advantages of different decoders, enhancing prosthetic limb control.
• This approach offers a more natural and reliable user experience for prosthetic limb users.