Electroencephalography based brain controlled grasp and lift

Abstract

The complexities and challenges of performing daily fundamental activities like getting dressed, answering a phone call, opening and closing the door, writing something down, or even consuming foods for patients who have lost their functionality of hands and arms due to neurological disability or amputations, is something anyone could never imagine. Our research paper serves to show the importance of restoring patients’ ability to do daily activities to increase their mobility and standard of living. In this paper, we have proposed an innovative, resilient and dynamic implementation of a grasp and lift technology that would accumulate brain signals in the form of waves to operate prosthetic limbs without the help of an external device and wires. We have decided to use Electroencephalography to reactivate the neuromuscular bypass with the help of an EEG device for obtaining brain signals that correspond to specific circumstances from the scalp surface area. We also have established models using Neural Networks that would monitor multimodal sensory activities which include object encounter, grasp, lift-off, replacement from the dataset and assist the users of this technology to operate the prostheses only by incorporating their brain signals. The artistry of the whole procedure incorporates substantial segments like signal acquisition and pre-processing of the signals into data, feature extraction, denoising etc. which later leads us to implement CNN and LSTM models. After implementing the models we obtained the accuracy of 90.11% and 74.44% from the CNN and LSTM model respectively. Throughout the implementation, there will be a differential boost in the accuracy level for each of the models. Therefore, our paper is an evidence of how EEG is considered to be a communication channel between prosthetic devices and the human brain. Furthermore, it intricately reveals the approach of grasp and lift technology through signal acquisition, processing, and implementation based on Electroencephalography.