Research On Wheelchair Robot Control Based On The Fusion Of EEG And Eyeball Signals

With the rapid development of Cloud Computing,Big Data and Artificial Intelligence,Natural Human-Computer Interaction has gradually integrated into human life and has an impact on human life.Natural Human-Computer Interaction mainly solve the problem of interaction between people and computers,at present,there are many kinds of interaction between people and computers,how to interact more naturally and seamlessly,has always been the direction and purpose of the researcher's research.The acquisition and application of EEGs and eye-movement signal is the research direction of many researchers in the study of natural human-computer interaction.EEGs and eye-movement signals are collected from human beings and are the typical way for the characteristic signals of human beings to be applied to the external computer equipment.For the variety of characteristic information,it is necessary to carry out reasonable information processing for rational use,multiinformation fusion is one of the most effective methods to study information processing.At present,the application of EEG,eye-movement signals and multi-information fusion are all beginning to serve humanity and affect the healthy lifestyle of human beings.This article is based on the background of Natural Human-Computer Interaction,The EEGs and eye-movement signals were collected to analyze and generate decision signals.The method of multi-information fusion is designed for two kinds of decision signals.To serve the signal of fusion in human life,achieve the goal of natural human computer interaction.Based on the general electric wheelchair,the platform of wheelchair robot system is designed,and lays the foundation for improving the quality of life of the target population.The research contents of this paper mainly include the following aspects:(1)The physiological mechanism of EEG was studied and the hardware structure of EEG collector is studied,and the corresponding collector is designed.According to the actual situation of EEG collecting,we put forward the EEG collection specification.The feature extraction and pattern classification of EEG were studied.Proposes an improved Common Spatial Pattern and Deep Belief Network which is used in the recognition and classification of EEGs,with the improved method of Common Spatial Pattern a variety of characteristic signal are extracted from EEG signals.The extracted characteristic signals are identified and classified by the Deep Belief Network realized the purpose of a variety of EEGs classification.Through experiment contrast,the classification accuracy of the improved Common Spatial Pattern and Deep Belief Network is higher than that of traditional EEGs.It provides a research idea for the identification and classification of various EEGs in the future study.(2)The AdaBoost detection algorithm is used to detect face and ensure the accuracy and real time of face detection.In the process of face detection,adopt the rectangle features that suitable for face detection and the training of cascade classifier is carried out by using these rectangular features.It meets the requirement of face detection,and lays the foundation for the next step.(3)On the basis of face detection,AdaBoost detection algorithm is used to locate human eyes.Because of the difference between human eyes and facial features,a new rectangular feature template is designed.After the human eye is determined,the eye direction is determined by the difference between the pupil center and the Angle of the eye.The determination of the center of the pupil adopts the pixel binarization processing,and the position of the eye angle is gradually determined according to the center position of the pupil.By using the relative position distance of the two,we can determine the position of the vision projection onto the computer screen,thus obtaining the eye-movement signal.(4)As for EEGs and eye-movement signals.According to the principle of multiinformation fusion in information processing,we designed the corresponding multiinformation fusion method to obtain the decision signal of the fusion.In order to make the decision signal serve the human race.After studying the signal transmission method and the control process,the multi-information source controlled wheelchair robot system is designed.The final experiment shows that: The wheelchair robot with multi-information source control has better Natural Human-Computer Interaction ability.The fusion of EEGs and eyemovement signals improves the control accuracy of wheelchair robot,enhanced the determination of Natural Human-Computer Interaction to serve human life.