A brain-computer interface integrated with virtual reality and robotic exoskeletons for enhanced visual and kinaesthetic stimuli

Brain-computer interfaces (BCI) allow the direct control of robotic devices for neurorehabilitation and measure brain activity patterns following the user’s intent. In the past two decades, the use of non-invasive techniques such as electroencephalography and motor imagery in BCI has gained traction. However, many of the mechanisms that drive the proficiency of humans in eliciting discernible signals for BCI remains unestablished. The main objective of this thesis is to explore and assess what improvements can be made for an integrated BCI-robotic system for hand rehabilitation. Chapter 2 presents a systematic review of BCI-hand robot systems developed from 2010 to late 2019 in terms of their technical and clinical reports. Around 30 studies were identified as eligible for review and among these, 19 were still in their prototype or pre-clinical stages of development. A degree of inferiority was observed from these systems in providing the necessary visual and kinaesthetic stimuli during motor imagery BCI training. Chapter 3 discusses the theoretical background to arrive at a hypothesis that an enhanced visual and kinaesthetic stimulus, through a virtual reality (VR) game environment and a robotic hand exoskeleton, will improve motor imagery BCI performance in terms of online classification accuracy, class prediction probabilities, and electroencephalography signals. Chapters 4 and 5 focus on designing, developing, integrating, and testing a BCI-VR-robot prototype to address the research aims. Chapter 6 tests the hypothesis by performing a motor imagery BCI paradigm self-experiment with an enhanced visual and kinaesthetic stimulus against a control. A significant increase (p = 0.0422) in classification accuracies is reported among groups with enhanced visual stimulus through VR versus those without. Six out of eight sessions among the VR groups have a median of class probability values exceeding a pre-set threshold value of 0.6. Finally, the thesis concludes in Chapter 7 with a general discussion on how these findings could suggest the role of new and emerging technologies such as VR and robotics in advancing BCI-robotic systems and how the contributions of this work may help improve the usability and accessibility of such systems, not only in rehabilitation but also in skills learning and education.