Discrete-Target Prosthesis Control Using Uncertainty-Aware Classification for Smooth and Efficient Gross Arm...

Area of Science:

• Biomedical Engineering
• Rehabilitation Robotics
• Human-Computer Interaction

Background:

• Continuous control for prosthetic arms is prone to fluctuations from signal variations.
• Discrete target pose control offers potential but is under-explored and susceptible to misclassification errors.

Purpose of the Study:

• To develop and evaluate a novel Uncertainty-Aware Discrete-Target Prosthesis Control (UA-DPC) approach.
• To mitigate unintended pose switches and enhance prosthesis movement smoothness and efficiency.

Main Methods:

• Implemented an uncertainty-aware classification scheme to minimize misclassification impact.
• Integrated real-time trajectory planning adjusting motion based on quantified uncertainty.
• Conducted human-in-the-loop experiments in VR with 12 participants controlling a transhumeral prosthesis.

Main Results:

• UA-DPC demonstrated superior performance compared to traditional discrete and continuous control methods.
• Achieved more efficient task completion with significantly fewer misclassification instances.
• Resulted in smoother movements of the residual limb and prosthesis.

Conclusions:

• The proposed UA-DPC approach offers a more efficient and stable control strategy for prosthetic arms.
• Addressing classification uncertainty is key to improving discrete-target prosthesis control.
• UA-DPC facilitates smoother and more reliable prosthetic limb function.