A Real-Time Motor Imagery Decoding Paradigm for Robot Manipulation

Spinal Cord Injury (SCI) has become a severe concern worldwide, with growing mortality rates anticipated to become the third highest reason behind death in the last decade. One approach to enable communication between individuals with SCI and machines is identifying and interpreting brain activity patterns. This paper proposes a deep learning framework combining Brain-Computer Interfacing and upper limb prosthesis as a rehabilitation tool for SCI patients, where after successful feature extraction by fast Fourier transform, different attempted arm and hand movements have been detected from the motor imagery signal obtained from an electroencephalography sensor. The gated recurrent unit-based classification results with 97% accuracy have been compared with the state-of-the-art techniques. Also, the robot’s desired movement detection performance has been compared with existing robot control strategies, proving that the proposed approach has remained consistently better than the existing competitors making it suitable for the healthcare industry.