Analysis Of Path Preference Identification Of Intelligent Wheelchair Users Based On Physiological Signals

With the increase of aging population,chronic disease patients and accidental injury events,and the decrease of new population,China's population is facing the challenge of aging.More and more people who are disabled and semi disabled,and who are mentally retarded and semi retarded,have a huge burden on their families and society.In order to improve and improve the quality of life and freedom of people who are lack of mobility,intelligent wheelchairs are more and more valued and supported by people from all walks of life.However,in the process of completing a task,the traditional auxiliary robot does not focus on the subjective feelings of human emotions and wishes,but mechanically "transports" users like transporting goods.Therefore,the purpose of this project is to establish an emotion recognition model,which aims to identify the emotion generated by human and robot in the interaction process,and set up a behavior criterion for robot based on human emotion,so that robot can consider the user's preference and choose its own "like" path in the process of performing tasks,improve user comfort,and realize human-computer interaction Harmony.Aiming at the human-computer interaction situation,this paper takes the psychological needs of human beings as the research focus.By building the emotion inducing platform and physiological signal detection platform,combined with the immersive virtual reality technology,12 groups of virtual reality experimental scenes are designed under the human-computer interaction situation.When users use the wheelchair for automatic driving,the distance between human beings and obstacles and the number of obstacles are respectively The number of obstacles,the shape of obstacles and the width of channels which affect the user's path selection are studied as emotional variables.In this study,a two-dimensional emotional model is selected to distinguish the types of emotions.The model divides emotions into two vectors: potency(pleasant or unpleasant)and arousal(intensity of emotions).In order to get more accurate experimental data,we screened and trained all the target subjects before the experiment.During the experiment,the emotion of the subjects was induced by virtual reality scene,and the physiological signals of the subjects were detected at the same time.We analyzed the sample data obtained in the experiment,and found the direct relationship between physiological signals and emotions.Through the relationship between the two,we builta convolutional neural network emotion recognition model based on text classification,and trained the model with 65 training set samples.The training results showed that the training effect curve fit well,and the model achieved precision after 100 iterations Requirement.The model is tested with 40 test samples,and the overall recognition rate is 92.5%.The results show that the trained model can accurately predict the emotion represented by physiological signal samples.