Controlling Virtual Simulation Using Brain-Machine Interfaces

Abstract

This project aims to introduce an improved application of brain-machine interfaces (BMI) to controlling virtual-reality simulations. Although frequently seen in science fiction, combining brain-machine interfaces with virtual-reality is a relatively unexplored field. In our setup, a test subject navigated around a virtual maze using electroencephalography (EEG) signals produced from imagining the use of his left or right hand. Using the non-linear filter algorithm for motor imagery classification, the test subject was able to navigate about the virtual-reality maze in the Illinois Simulator Laboratory. Experiments show that the test subject is able to follow a path in a 4x4 maze with error rates ranging between 0 to 8.70%.