A Novel Framework to Facilitate User Preferred Tuning for a Robotic Knee Prosthesis

Area of Science:

• Biomedical Engineering
• Rehabilitation Engineering
• Human-Robot Interaction

Background:

• Personalized control tuning is crucial for robotic prosthesis adoption.
• Existing automatic tuning methods often neglect user preferences.
• User-centered design is needed to improve prosthetic device personalization.

Purpose of the Study:

• To propose and evaluate a novel framework for tuning robotic knee prosthesis control.
• To enable user-preferred knee kinematics through a personalized tuning process.
• To investigate user ability to identify preferred prosthetic control profiles.

Main Methods:

• Developed a framework with a User-Controlled Interface for selecting preferred knee kinematics.
• Implemented a reinforcement learning algorithm to tune high-dimensional prosthesis control parameters.
• Conducted evaluations of framework performance, usability, and user preference identification through blinded studies.

Main Results:

• The framework effectively tuned 12 robotic knee prosthesis control parameters to meet user-selected kinematics.
• Users could accurately and consistently identify their preferred prosthetic control knee profile in blinded tests.
• Preliminary gait biomechanics analysis showed no significant difference between preferred control and normative control.

Conclusions:

• The developed framework successfully personalizes robotic knee prosthesis control based on user preferences.
• User preference plays a significant role in the adoptability and effectiveness of robotic prostheses.
• This approach holds potential for future clinical and home use of personalized prosthetic devices.