Control of a virtual reality environment for upper limb movement using a motor imagery-based brain computer interface

Abstract

The following project is presented as a thesis proposal for the Master of Science in Computer Science (MCC-i): the design, implementation and evaluation of a virtual reality (VR) environment controlled using a Motor Imagery based Brain Computer Interface (BCI). BCIs enhance the effective communication and interaction between humans and computers. Such systems are increasingly prevalent in diverse applications, including education, entertainment, and health. The aim of this system is to reate a rehabilitation environment for upper limb motor recovery, in the form of a VR game. In the system, the user performs left and right arm MI, which is detected using machine learning algorithms to perform a movement and ability within the virtual environment. The system was evaluated with five healthy participants in one experimental session each. Each experimental session consisted of a training and an evaluation routine, in which the participants were asked to imagine each MI task andomly to gather training data and then, with the trained classification algorithm, the users were evaluated by playing the game in which they needed to perform the correct ability in the game to attack an enemy. The classification algorithms showed a ranged accuracy of 39.6% to 68.6%, with an average of 51.9%; the evaluation accuracy ranged from 38.5% to 76.9%, with an average of 58.5%. A User Experience Survey was applied to the participants, obtaining positive results and feedback on game improvements.