A method to determine the optimal features for control of a powered lower-limb prostheses

Area of Science:

• Biomedical Engineering
• Rehabilitation Robotics
• Prosthetics and Orthotics

Background:

• Lower-limb prostheses are advancing with increased computational power and sensing capabilities.
• Control strategies for prostheses often involve a trade-off between extrinsic (e.g., electromyography) and intrinsic (internal sensor) modalities.
• Identifying effective control features is crucial for improving prosthetic function and user experience.

Purpose of the Study:

• To explore the trade-offs between extrinsic and intrinsic control modalities for lower-limb prostheses.
• To develop a method for identifying features that discriminate between locomotion modes.
• To assess the applicability of these features for different amputation types and for unknown terrain transitions.

Main Methods:

• Collected locomotion data from two trans-femoral and one trans-tibial amputee participants on varied terrain.
• Utilized a large set of features derived from both electromyographic (extrinsic) and internal sensors (intrinsic).
• Developed and applied a method to identify features that perfectly discriminate between locomotion modes, independent of amputation level.

Main Results:

• Successfully identified specific features that effectively discriminate between different locomotion modes during walking.
• The identified discriminatory features were found to be independent of the type of lower-limb amputation.
• The study provides insights into selecting optimal sensor/feature combinations for prosthetic control, particularly for challenging transitions.

Conclusions:

• A novel method can identify key features for prosthetic limb control that are transferable across amputation types.
• This research contributes to optimizing sensor selection for intrinsic control in lower-limb prosthetics.
• The findings support the development of more intuitive and adaptable prosthetic leg control systems.