A two-class self-paced BCI to control a robot in four directions

Area of Science:

• Neuroscience
• Robotics
• Human-Computer Interaction

Background:

• Brain-computer interfaces (BCIs) offer potential for controlling external devices.
• Electroencephalography (EEG) is a common modality for BCI signal acquisition.
• Self-paced (asynchronous) BCIs are desirable for naturalistic control.

Purpose of the Study:

• To propose and evaluate an EEG-based, self-paced BCI for mobile robot navigation.
• To investigate the efficacy of using two mental tasks for simplified BCI control.
• To assess the BCI's performance in both simulated and real-world environments.

Main Methods:

• An asynchronous BCI system was developed using electroencephalographic (EEG) analysis.
• Four navigation commands (forward, back, left, right) were implemented.
• An audio-cued interface utilized two mental tasks: relaxed state and right-hand movement imagination.
• Experiments involved four healthy subjects, including training in a virtual environment and operation of a real robot.

Main Results:

• Three out of four subjects successfully controlled the mobile robot in a real environment.
• The BCI demonstrated effective robot navigation capabilities.
• The system showed promise for individuals with lower BCI performance.

Conclusions:

• The proposed EEG-based, self-paced BCI is a viable method for mobile robot control.
• Simplified BCI paradigms using few mental tasks can achieve functional control.
• The interface is robust enough to allow operation in real-world scenarios, even for novice users.