A complex approach to increasing the quality of the biocontrol of robotic wheelchairs

The continuing challenge of improving the quality of the biocontrol of robotic wheelchairs is addressed and a solution consisting of an integrated approach based on the combined use of gaze control and neurocontrol methods is proposed; the stages in implementing this approach are considered. The problems arising in creating gaze-controlled robotic vehicles are discussed and challenges occurring in the implementation of biocontrol and ways to solve them are analyzed. A method for selecting the most effective electroencephalography leads, based on the use of an international signal database, was developed. A convolutional neural network architecture was developed for the joint classification of eye-tracking and brain signals. Results from a study of neural networks are presented, these demonstrating a maximum prediction accuracy using the pre-trained InceptionV3 model.