"MindDrive" - Personalized Model Using EEG-Based Brain-Computer Interface for Wheelchair Control
Aanya Krishna
21/07/2026
Our project addresses several challenges of traditional wheelchairs and previous BCI-based prototypes to make the system more usable. Current solutions are often based on high channel EEG configurations or deep learning models with high computation and processing time. This approach employs an 8-channel EEG headset that is easy to set up and does not require extensive preprocessing. A user-specific classifier was chosen for its accuracy and efficiency, enabling robust user intention classification with limited data. The algorithm identifies user intents like "left," "right," "move" or "stop" and sends them to the motor controller of the wheelchair in a virtual environment. The user-specific model had an accuracy of >90% and the user successfully navigated in the virtual environment in 120 seconds compared to 97 seconds when using a keyboard. By removing the need for manual or vocal controls, the system is fully brain-controlled, and is particularly beneficial for people with motor or speech disabilities. The project is also low-cost and portable to facilitate implementation.