EEG Controlled Robotic Arm System for Motor Command Reception in Prosthetic Limbs

Individuals without the use of a limb face significant challenges in interacting with their surroundings. These unique challenges are often remedied with the use of a prosthetic limb. However, prosthetics provide limited movement options, lack natural neural control from the user, and require specific maintenance guidelines. This research endeavors to enhance the usability of artificial prosthetics by enabling seamless interaction between the user's brain and the prosthetic limbs. This will be achieved by contributing to the collection of human brain signals and development of machine learning (ML) models, such as convolutional neural network (CNN), that can decode and identify limb movements. Brain signals were collected using an electroencephalogram (EEG) cap and fed into an ML model to identify motor commands, which were then used to control a robotic arm representing the prosthetic limb. Image processing techniques were used to provide real-time input for the ML model, which was trained using EEG signals from 9 subjects. The model achieved approximately 70% accuracy in interpreting motor commands to control the robotic arm with a low latency of 4 ms.