| The traditional mechanized human-computer interaction of service robots is difficult to meet people's growing needs for a better life.Nowadays,emotional interaction has become an important research direction in the field of robotics.Merabin's law points out that about55% of emotional information in people's daily communication is revealed through facial expressions,so the research on facial expression recognition technology is of great significance to the establishment of friendly human-computer emotional interaction.Aiming at the fact that most service robots lack the ability of emotion recognition and emotional interaction based on facial expressions,this paper conducts in-depth research on expression recognition technology and its application to service robots.The main research work is as follows:First of all,in view of the lack of theoretical model support for the robot's cognitive calculation of facial expressions,the facial expression description model is explained through the physiological basis of facial expressions and the classification description model of facial expressions,which combines human cognitive calculations with The analogy of the principle of visual cognition is transferred to the robot,and on this basis,a theoretical model of the robot's cognitive calculation of facial expressions is established.Secondly,in view of the lack of system scheme design of facial expression recognition technology based on deep learning,the system scheme design is carried out in order from the three parts of deep learning model construction of facial expression recognition algorithm,model training and model reasoning.Partially select the basic framework of the deep neural network model and the tools for constructing the deep learning framework in the model building;select the appropriate data set for preprocessing in the model training part,and explain the mechanism of model training;build the image in the model reasoning part On the acquisition platform,the image is preprocessed by face detection,face image cropping and normalization,and after inputting the trained model for reasoning,the design of the expression recognition system is completed.Then,for the common problems of high-precision deep learning model parameters and calculations,large memory usage,and long training time,a lightweight deep convolutional neural network is used to build a robot facial expression recognition algorithm model.First,build different expression recognition algorithm models based on the fine-tuning of the classic convolutional neural network structure,and then design and build a lightweight deep convolutional neural network model based on the advantages of various network models,and compare the results of the model training through related parameter indicators Through analysis,the optimal expression recognition algorithm model that can be used in the robot expression recognition system is obtained.Finally,in view of the lack of application of facial expression recognition technology on service robots,an emotional interaction system for robots is designed and implemented.Under the premise of the application scenario in the shopping mall,the UI interaction interface of the backend of the service robot expression recognition system was developed,and the human-machine emotional interaction was realized by combining the voice interaction strategy.The research results in this paper show that the established robot facial expression cognitive computing theoretical model provides a practical basis for realizing facial expression recognition.The designed facial expression recognition algorithm model can reduce the amount of calculation and parameters while maintaining high recognition accuracy.The facial expression recognition system can accurately recognize and display multiple facial expression recognition results,and finally the service robot can perform facial expression recognition in real time and provide emotional interaction. |
PDF Full Text Request