Systematic Literature Review: Implemented Motor Imagery Recognition Methods

Initiating a physical movement triggers a thinking process in the brain. This thinking process leads to the creation of electroencephalography (EEG) signals related to the body's movements. This process—then known as motor imagery (MI)—is one of the fields of brain-computer interface (BCI) or brain waves that utilizes EEG. In another definition, MI refers to human beings imagining themselves performing motor movements or moving limbs. Numerous high-quality studies on motor imagery recognition have been published due to its many potential benefits. However, a significant portion of them do not put it into implementation. Based on our study, Neural Networks (NN) is the most used for recognizing implemented motor imagery. Private data are crucial for motor imagery applications because these data allow for better control and analysis. This study also found that motor imagery is being extensively used in tools like robotic/virtual hands or arms, movements of humanoid robots (back, forth, and turning), and implementation to a skeleton robot that is often used for rehabilitation purposes. Based on the results for different utilizations, the accuracy is satisfactory. However, there is room for improvement in various specifications.