Continuous Description of Human 3D Motion Intent Through Switching Mechanism

Area of Science:

• Neuroscience
• Robotics
• Rehabilitation Engineering

Background:

• Early and active rehabilitation is crucial for post-stroke motor recovery and brain reorganization.
• Cable-based robots can assist in rehabilitation, but effective human-machine interfaces are needed to promote patient participation.

Purpose of the Study:

• To develop a novel method using electromyography (EMG) decoders to estimate voluntary motion intention for cooperative human-machine interfaces in robot-assisted rehabilitation.
• To improve the assistance for multi-joint complex tasks by developing a switching mechanism to decompose tasks into simpler motions.

Main Methods:

• A switching mechanism was developed to divide complex 3D tasks into simple motions.
• Linear time-invariant models were established for each simple motion, using processed muscle activations from six arm muscles as inputs and voluntary forces as outputs.
• Experiments with seven healthy participants involved model training, testing, and controller application using a multi-directional tracking task with visual feedback.

Main Results:

• The developed EMG decoder model, incorporating the switching mechanism, effectively recognized arm movement intentions.
• The system provided appropriate assistance to participants during rehabilitation exercises.
• The switching mechanism significantly improved both the accuracy of model estimation and the completeness of task execution for complex movements.

Conclusions:

• The proposed EMG-based method and switching mechanism offer a promising approach for enhancing robot-assisted stroke rehabilitation.
• This cooperative human-machine interface can promote greater patient participation and improve motor recovery outcomes.
• The system's ability to accurately decode intention and assist in complex tasks highlights its potential for clinical application.