Navigation in a virtual environment using multiclass motor imagery Brain-Computer Interface

Abstract
Virtual Reality is a useful platform for Brain-Computer Interface (BCI) users as it offers a relatively safe and cost-effective way for BCI users to train and familiarize themselves with using BCI in a virtual environment before using it in a real-world scenario. Hence this paper presents a pilot study of a virtual navigation task, where control signals from a synchronous multi-class motor imagery-based BCI (MI-BCI) is used by the subject to perform a navigation task in a 3D virtual environment, from a first-person perspective displayed on the computer screen. Preliminary results on one healthy subject showed that the MI-BCI was able to distinguish between 4 classes of motor imagery with an accuracy of about 67.5%, and the subject was able to navigate through the virtual environment in 87 trials in contrast to a theoretical minimum of 74 trials. Results from this study provide motivation to further investigate the potential of the MI-BCI in a larger-scale study, with the possibility of future clinical applications such as a training tool for users in BCI-based rehabilitation and other assistive technologies such as neural prosthetics or brain-controlled wheelchairs.