Research And Development Of Gesture Control Intelligent Nursing Bed Based On Deep Learning

At present,China has entered the period of rapid growth of population aging,the population structure is changing rapidly.With the growth of the age and the degeneration of the body function,the old people’s demand for nursing bed is increasing day by day,the function of nursing bed is perfected gradually.However,the control mode of the existing electric nursing bed mainly adopts the push-button type.Due to the poor eyesight of the elderly and other reasons,the push-button electric nursing bed has caused a lot of inconvenience to the self-care of the elderly.Adding more convenient man-machine interaction mode into the nursing bed is the key problem that needs to be solved in the field of nursing technology.In recent years,with the rapid development of deep learning theory,gesture recognition technology has made great breakthroughs,which greatly promotes the feasibility of gesture control scheme.The performance of embedded devices represented by raspberry PI is constantly improved,which makes it possible for complex deep learning model to run in real-time in embedded devices after compression.Therefore,based on the deep learning theory and embedded system,this paper optimizes the control mode and function of the existing electric nursing bed,thus solving the difficulties faced by the elderly in the process of self-care.The main research contents of this paper include:(1)This paper studied the SSD(Single Shot Multi Box Detector)framework in the target detection models,then this paper proposed and established an HD＿SSD gesture detection model for gesture recognition.This article first analyzes the application scenario of gesture recognition,and establish the gesture data set(Hand Control Dataset).According to the analysis of the data set,HD＿SSD(Hand Detection SSD)gesture detection model is proposed which combined with the Receptive Field(Receptive Field)theory and the FPN network(Feature Pyramid Networks),implements the rapid detection,accurate recognition of Hand gestures.(2)In this paper,raspberry PI 4B is used as the control chip of the intelligent nursing bed,and image acquisition is completed with a special camera,realizing the gesture control intelligent nursing bed scheme.In this paper,The HD＿SSD gesture detection model was compressed by using Int8 quantization,Uniform pruning and other operations,and the model’s running time in raspberry PI 4B was shortened from 1.35 s to 0.175 s,realizing the feasibility application of gesture interaction based on deep learning model in the field of nursing.(3)This paper completes the development of gesture control intelligent nursing bed from circuit design and software design.The development of gesture control intelligent nursing bed is completed through the design of control circuits such as electric push rod control,infrared temperature sensor control,liquid level sensor control,water pump photo-coupling isolation control,PTC heating tube PWM control and the software programming of control program.Finally,this paper verified the developed gesture control intelligent nursing bed through experiments.The results show that gesture interaction has practical feasibility in the field of nursing,which lays a foundation for the further industrialization of gesture recognition intelligent nursing bed.