Real-time optimization of an ellipsoidal trajectory orientation using muscle effort with Extremum Seeking Control

Area of Science:

• Robotics
• Control Systems
• Biomechanics

Background:

• Model-free optimization is crucial for real-time control in human-robot interaction.
• Extremum Seeking Control (ESC) offers a promising alternative to computationally intensive global optimizers like Biogeography-based Optimization (BBO).
• Previous studies utilized BBO for offline optimization, highlighting its limitations in real-time applications.

Purpose of the Study:

• To present and evaluate a real-time, model-free optimization approach for elliptical trajectories using ESC.
• To compare the performance of ESC against BBO in simulation using diverse arm models.
• To validate the ESC approach in experimental settings using a robotic exoskeleton and electromyography (EMG) for muscle activity analysis.

Main Methods:

• Implemented ESC as the real-time optimizer for elliptical trajectory orientation.
• Utilized a 4-degrees-of-freedom robot with impedance control for experimental validation.
• Measured muscle activations via electromyography (EMG) sensors during user interaction with the robot.
• Compared ESC performance against BBO using five distinct human arm models in simulations.

Main Results:

• ESC successfully optimized elliptical trajectory orientation in real-time, converging to optimal solutions within short trials (50s).
• The ESC approach demonstrated comparable performance to BBO across various arm models.
• Experimental results validated the feasibility of ESC for adaptive robotic exercise protocols.
• EMG data provided insights into muscle engagement, allowing for therapist-defined training objectives.

Conclusions:

• ESC provides an effective and computationally efficient model-free solution for real-time trajectory optimization in robotic rehabilitation.
• The developed framework enables personalized exercise protocols by adapting to user biomechanics and muscle activity.
• This approach holds potential for enhancing the efficacy and adaptability of robot-assisted physical therapy.