Using virtual reality to boost the effectiveness of brain-computer interface applications

Noninvasive brain-computer interface (BCI) uses data from electroencephalographic (EEG) sensors to train a model using machine learning and pattern detection algorithms to recognize certain patterns. These patterns correspond to a command to an actuator like wheels of a wheelchair, keyboard or mouse. Virtual reality is used to make a virtual environment (VE) where the users can control virtual objects. BCI generated commands can be used to control the virtual objects with a suitable interface. Unity 3D software provides the interface and the VE. The noninvasive nature of input collection makes the input prone to noise and error. The user needs considerable amount of training to learn to control the wheelchair efficiently. This training is done in a VE where the user controls a virtual wheelchair. The accuracy of wheelchair control before and after training in the VE is compared. It is observed that the user can control the physical wheelchair with better accuracy after training in the VE.