The SSHVEP Paradigm-Based Brain Controlled Method for Grasping Robot Using MVMD Combined CNN Model.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Steady-state visual evoked potentials (SSVEP) offer high data rates for brain-computer interfaces (BCIs) but struggle in dynamic settings.
• Existing SSVEP BCIs lack adaptability and require improved decoding accuracy for real-world applications.

Purpose of the Study:

• To develop an adaptive steady-state hybrid visual evoked potentials (SSHVEP) paradigm for improved brain-computer interaction.
• To enhance EEG decoding accuracy using a novel multivariate variational mode decomposition (MVMD) and convolutional neural network (CNN) approach.

Main Methods:

• Proposed a novel SSHVEP paradigm utilizing environmental grasping targets for enhanced subject-environment connection.
• Implemented an EEG decoding method combining MVMD for adaptive sub-band decomposition and CNN for target recognition.
• Conducted offline and online experiments with 18 subjects using a 9-target SSHVEP paradigm and a brain-controlled grasping robot.

Main Results:

• Offline accuracy reached 95.41 ± 2.70% with the SSHVEP paradigm, a 5.80% improvement over conventional methods.
• Online experiments with a brain-controlled grasping robot achieved an average accuracy of 93.21 ± 10.18%.
• The proposed method demonstrated significant improvements in decoding accuracy and adaptability.

Conclusions:

• The SSHVEP paradigm effectively enhances brain-computer interaction in dynamic environments.
• The MVMD combined with CNN algorithm significantly improves EEG decoding accuracy for BCIs.
• This research validates a robust and accurate brain-computer interaction method for assistive technologies.