AbstractMotor imagery Brain-Computer Interface (MI-BCI) enables bodyless communication by means of the imagination of body movements. Since its apparition, MI-BCI has been widely used in applications such as guiding a robotic prosthesis, or the navigation in games and virtual reality (VR) environments. Although psychological experiments, such as the Rubber Hand Illusion - RHI, suggest the human ability for creating body transfer illusions, MI-BCI only uses the imagination of real body parts as neurofeedback training and control commands. The present work studies and explores the inclusion of an imaginary third arm as a part of the control commands for MI-BCI systems. It also compares the effectiveness of using the conventional arrows and fixation cross as training step (Graz condition) against realistic human hands performing the corresponding tasks from a first-person perspective (Hands condition); both conditions wearing a VR headset. Ten healthy subjects participated in a two-session EEG experiment involving open-close hand tasks, including a third arm that comes out from the chest. The EEG analysis shows a strong power decrease in the sensory-motor areas for the third arm task in both training conditions. Such activity is significantly stronger for Hands than Graz condition, suggesting that the realistic scenario can reduce the abstractness of the third arm and improve the generation of motor imagery signals. The cognitive load is also assessed both by NASA-TLX and Task Load index.
Toward Expedited Impedance Tuning of a Robotic Prosthesis for Personalized Gait Assistance by Reinforcement Learning Control
