EEG-based BMI using Multi-Class Motor Imagery for Bionic Arm

One of the rehabilitation methods developed to help people with physical disabilities, especially patients with disabilities of the upper-limb arm, to do Activities of Daily Living (ADL) is the Bionic arm. The commonly used bionic arm control method is utilizing residual muscle motor neurons. However, the problem arises in patients with upper elbow amputation, limited residual muscle, muscle cells damaged or not growing again which causes the arm to not be able to give enough signals. The electrical activity of the brain recorded using electroencephalography (EEG) can be utilized for bionic arm control. The movement of grasping an object to fulfill ADL is a complex movement, so a system that is able to overcome these needs is needed. The Brain-Machine Interface (BMI) with Motor Imagery (MI) on a bionic arm multiclass control system based on Fast Fourier Transform (FFT) and Support Vector Machine (SVM) is described and discussed in this article. The system built can handle complex movements grasping various objects and evaluated for use with long pauses. The experiment results show good SVM-FFT performance when implemented on a real bionic arm. The use of Additional Movement (AM) increases classification success by up to 20%. On the other hand, this system is able to handle usage even though it has been rested in a few minutes.