The Steady State Visually Evoked Potential (SSVEP) phenomenon, where brain activity responds to flashing lights, enables hands-free machine control. This project explores the development of a Brain-Computer Interface (BCI) using Convolutional Neural Networks (CNNs) to detect SSVEPs, facilitating hands-free interaction for astronauts and pilots. By placing EEG sensors over the occipital lobe—responsible for visual processing—we can observe neuronal activity as users focus on one of six flashing lights, each emitting a unique frequency. Neurons in the occipital lobe synchronize with the chosen light, producing a spike in brain wave activity matching the light’s frequency.
We trained a CNN on EEG data from over 45 participants, resulting in a generalized model with 85% accuracy for familiar users and 70% for new users. Using transfer learning, the model adapts to specific individuals, achieving over 90% accuracy with minimal additional training. In real-time applications, a user wearing an EEG headset focuses on a flashing light in a layout resembling a controller, with each light corresponding to an action. The CNN detects the user's focus within two seconds, enabling hands-free control of machines in environments where manual control is impractical. This technology offers a reliable, albeit unusual, alternative control method for professionals in critical scenarios