Brain-machine interface control of a manipulator using small-world neural network and shared control strategy

Area of Science:

• Neuroscience
• Robotics
• Biomedical Engineering

Background:

• Brain-computer interface (BCI) research aims to improve brain signal resolution and external device control.
• Controlling external devices via non-invasive brain signals remains a significant challenge.

Purpose of the Study:

• To develop and validate a non-invasive brain-actuated manipulator system.
• To establish a paradigm for motion control of a serial manipulator using motor imagery and shared control.

Main Methods:

• Utilized component selection, spatial filtering, and classification of motor imagery.
• Employed a small-world neural network (SWNN) for classifying five brain states, outperforming RBF, SMN, and multi-SVM classifiers.
• Implemented a shared control method with two control patterns for expanded BMI software control.

Main Results:

• The proposed SWNN classifier demonstrated a 3.83% improvement over other tested classifiers.
• The developed BMI system successfully achieved motion control of a manipulator throughout its workspace.
• Experimental results confirmed the feasibility of the proposed BMI method for 3D manipulator control using electroencephalography (EEG) during motor imagery.

Conclusions:

• The study confirms the efficacy of the proposed non-invasive BMI for 3D manipulator motion control.
• The integration of motor imagery and shared control offers a promising direction for advanced BCI applications.
• This research advances the capabilities of brain-actuated robotic systems.