An Automatic Approach to Control Wheelchair Movement for Rehabilitation Using Electroencephalogram

An Automatic Approach to Control Wheelchair Movement for Rehabilitation Using Electroencephalogram

Darunjeet Bag, Ahona Ghosh, Sriparna Saha
Copyright: © 2023 |Pages: 22
DOI: 10.4018/978-1-6684-5381-0.ch007
OnDemand:
(Individual Chapters)
Available
$37.50
Current Special Offers
No Current Special Offers
TOTAL SAVINGS: $37.50

Abstract

Modern science and technology development enables humans to control devices using their brains. brain computer interface (BCI) system using EEG (Electroencephalogram) is a non-invasive method that uses brain signals to control robots. Brain control robots can assist people with disabilities to improve their livelihood. In our proposed work to control wheelchair movement for rehabilitation using EEG, electrodes are first placed on the subject's scalp to acquire brain signal activity. After that, the signals got filtered using fast fourier transformation (FFT) method, and then features got extracted using power spectral density (PSD). Random forest is used for the classification of wheelchair movement. For this purpose, a publicly available dataset from Kaggle is used, and an average accuracy of 96.79 is achieved. The proposed architecture has outperformed all the existing ones in its concerned domain; thus, it is suitable, cost-effective, and flexible for the users, which also helps maintain user privacy.
Chapter Preview
Top

Various researchers have researched our proposed framework (Ghosh & Saha, 2020; Saha & Ghosh, 2019). Here we have discussed some of the related works. Edla et al. (2018) use Neurosky Mindwave mobile as an EEG recording device. It is a non-invasive device that connects wirelessly with a computer or mobile using Bluetooth. This device consists of two electrodes placed on the forehead at the Fp1 location according to the 10-20 system and clip-on-ear as a reference electrode, respectively. They intend to classify two different states of the human mind: concentration and meditation. First, raw EEG signals of 8 different frequency bands are collected from the Fp1 channel: low gamma, high gamma, high alpha, low alpha, high beta, low beta, delta, and beta. After that, these raw EEG signals are preprocessed. For preprocessing, a single chip is presented in that Neurosky Mindwave device. Then useful features are extracted from that signal. First, the statistical tool is used for signal features, such as mean, standard deviation, and difference of the highest and the lowest value. After that, classification is done using a random forest machine-learning algorithm. It is an ensemble learning approach which applies a random amount of decision trees to develop the final prediction. They achieved an accuracy of 75% on predicting the class of the mental state.

Complete Chapter List

Search this Book:
Reset