EEG-based Control of Ground Robot via Brain-Computer Interface

In this thesis, a sophisticated EEG-based framework for real-time robotic control is proposed. The EEG signals are captured through the brain mechanism of inner speech, or ”the voice inside one’s head,” which serves as an experimental paradigm for EEG signal acquisition. A comprehensive brain-computer interface (BCI) is developed, consisting of different stages. These stages include the acquisition of EEG signals, preprocessing techniques to filter and enhance the signals, and robust classification algorithms to interpret the signals into corresponding commands. The final stage involves the integration of these commands into the robot’s control system, enabling movement control. This framework aims to demonstrate the feasibility and efficiency of using inner speech based EEG signals for intuitive and effective robotic manipulation. The four different commands for the robot to move are left, right, forward and backward. The architecture is tested both in simulation and real measurement environment. The empirical results provided a feasible proof of concept, demonstrating that the robot can be controlled via EEG signals, with a bias towards the third classification category. This category corresponds to the forward motion, achieving an accuracy rate of approximately 40%.