Extremum Seeking Control for Model-Free Auto-Tuning of Powered Prosthetic Legs

Area of Science:

• Robotics
• Control Systems
• Biomedical Engineering

Background:

• Manual tuning of prosthetic leg control gains is time-consuming and iterative.
• Balancing prosthetic leg tracking performance with user comfort is a key challenge.
• Existing methods lack real-time adaptation to varying walking conditions.

Purpose of the Study:

• To develop an extremum seeking controller (ESC) for simultaneous tuning of prosthetic leg feedback control gains.
• To create a convex objective function that integrates tracking error and user comfort.
• To enable real-time adaptation of prosthetic leg control for diverse walking speeds.

Main Methods:

• Development of a convex objective function incorporating tracking error and user comfort.
• Theoretical analysis of objective function properties and stability using averaging and singular perturbation.
• Real-time minimization of the objective function by ESC to tune proportional gains.
• Verification through benchtop and walking experiments.

Main Results:

• The ESC effectively tunes proportional gains for knee and ankle joints simultaneously.
• The controller demonstrates real-time adaptation to changes in walking speed.
• Stability of the closed-loop system with ESC is theoretically proven.
• Experimental results validate the ESC's effectiveness across various walking speeds.

Conclusions:

• The proposed ESC offers an automated and efficient solution for tuning prosthetic leg controllers.
• Real-time adaptation capability enhances prosthetic leg performance and user experience.
• The ESC provides a robust method for optimizing prosthetic leg control for dynamic environments.