Balancing a simulated inverted pendulum through motor imagery: an EEG-based real-time control paradigm

Area of Science:

• Neuroscience
• Robotics
• Biomedical Engineering

Background:

• Real-time brain-computer interfaces (BCIs) are crucial for controlling dynamic systems.
• Controlling unstable devices with BCIs presents significant challenges.

Purpose of the Study:

• To develop and evaluate a real-time feedback BCI paradigm.
• To enable control of an inverted pendulum on a cart (IPC) using SMRs.

Main Methods:

• Utilized 15 active scalp electrodes to record sensorimotor rhythms (SMRs).
• Applied Common Spatial Pattern (CSP) for spatial feature extraction.
• Employed Linear Discriminant Analysis (LDA) for translating neural patterns into control commands.

Main Results:

• Five subjects successfully balanced the simulated inverted pendulum for over 35 seconds.
• Demonstrated the capability of BCIs to control nonlinear, unstable devices in real-time.
• Showcased the feasibility of continuous control for practical BCI applications.

Conclusions:

• BCIs can effectively control complex, unstable robotic systems.
• The developed real-time BCI paradigm shows promise for real-life applications and generalization.
• This research advances the field of BCI control for dynamic environments.