A Greedy Assist-as-Needed Controller for Upper Limb Rehabilitation.

Area of Science:

• Robotics
• Neurorehabilitation
• Biomedical Engineering

Background:

• Active patient participation enhances robotic rehabilitation outcomes.
• Existing assist-as-needed (AAN) controllers do not account for individual motor function heterogeneity.
• Improved voluntary effort is crucial for effective rehabilitation training.

Purpose of the Study:

• To design a novel greedy assist-as-needed (GAAN) controller for upper limb robotic rehabilitation.
• To address the heterogeneity of motor function capabilities in neurologically impaired subjects.
• To enhance voluntary engagement and active participation during rehabilitation exercises.

Main Methods:

• Developed a GAAN control paradigm incorporating a baseline controller and a Gaussian Radial Basis Function (RBF) network.
• The RBF network models subject functional capability and provides task challenges.
• Implemented a greedy strategy for updating RBF network weights to track subject effort and a challenge modification algorithm.

Main Results:

• The GAAN controller successfully modeled individual subject capabilities and adjusted task difficulty.
• Experiments with 12 neurologically impaired subjects demonstrated the controller's feasibility.
• The GAAN controller showed significant potential in promoting voluntary engagement during training.

Conclusions:

• The proposed GAAN controller effectively addresses motor function heterogeneity in robotic rehabilitation.
• GAAN enhances subjects' active participation and voluntary effort.
• This approach holds promise for improving rehabilitation outcomes in neurologically impaired individuals.