EEG Classifier Using Wavelet Scattering Transform-Based Features and Deep Learning for Wheelchair Steering

A Brain-controlled wheelchair functions via a brain-computer interface (BCI), an electroencephalogram (EEG) worn on the patient's forehead to detect neural activity. Classifying human thoughts still needs a significant effort to achieve a reliable BCI This paper presents the design and testing of a mind-based control of a wheelchair for physically challenged users. The design of this system is focused on the acquisition of the EEG brain signals to create control of the wheelchair. All the laboratory measurements of human brain activities have been recorded using human control commands of wheelchair navigation. The received data are used to design a human brain-based control mechanism using wavelet scattering and deep learning techniques. After testing the designed deep learning algorithm, the overall classification accuracy was 95% to distinguish between four commands of the wheelchair (Forward, Stop, Left, and Right).