| Autor: |
Petrunina, Elena, Bayramov, Elmin, Pecherskij, Denis |
| Předmět: |
|
| Zdroj: |
Proceedings of the International Multidisciplinary Scientific GeoConference SGEM; 2023, Vol. 23, p81-88, 8p |
| Abstrakt: |
The research on the development of new robot control systems is currently underway. The interest in this area stems from the need for practical, user-friendly means of transport adapted to people with physical disabilities. The application of these devices will allow disabled people to manipulate external devices using EEG of both brain activity and eye movement. It is still possible for people with disabilities to partially apply their gaze in order to control and communicate with an assistive device. However, the application of eyetracker - based interfaces technology encounters the problem of involuntary eye movements leading to unwanted item selection (the Midas-Touch problem). The following issue can be addressed by the development of multimodal and hybrid management interfaces. Brain-computer interfaces implement translation of brain activity patterns into commands designed to control interactive applications, with recognition of motor imagination patterns. Thus, this study proposes a multimodal architecture for wheelchair gaze-control system for people with mobility impairments using gaze control and intention confirmation technology using brain-computer interfaces. In this study, a hybrid model was proposed to classify EEG motor imagery signals and eye tracker signals to implement a control system with a neural network architecture consisting of pre-trained convolutional neural network and gated recurrent unit. The performance of the adapted approach is determined using a multiclass imaginary motion dataset and the corresponding swipes and classification results. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
