Research On PVDF Sensor Array For Sense Of Motion

Gesture recognition,as an important topic in computer science and language technology,has two main implementations.One is machine vision-based gesture recognition,which is the currently mainstream of gesture recognition technology now.However,this way is highly dependent on the external environment.It is required that there is no occlusion between the user and the imaging device,so the application scenario is limited.The other one is the sensor-based gesture recognition,which is mainly realized by inertial sensors,piezoresistive sensors,etc.,such as the data gloves.However,this device would restrain the user's hand.In addition,the surface electromyography(sEMG)sensor is also the focus of research.These sensors have high requirements for wearing and the experience while using is poor.For the above situation,this thesis designed a sensor array for sense of the wrist motion by using the PVDF(Polyvinylidene Fluorid)piezoelectric film,and it can carry out finger motion recognition through the detection of the wrist motion signals.The main work content and innovations of this paper have the following three aspects:1.The non-invasive PVDF sensor array for sense of wrist motion was designed and fabricated.This thesis presents a PVDF sensor unit with the curved elastic backing and the integrated signal preprocessing circuit.Then,by analyzing the physiological of the hand,the scientific sensor array arrangement is designed.Finally,based on STM32 microprocessor,the real-time gathering and transmission system of the lower computer is constructed.And the signal acquisition experiments show that the sensor array can collect and transmit the wrist motion signals,and the collected signals of different finger movements have visible differences,which is enough for finger motion recognition as a difference sample.2.The database of finger motion signals has been established.This thesis builds a set of real-time signal reading and display system of the host computer on Windows.Then,according to the analysis of the physiology of the hand,the coding rule of finger action is proposed.Finally,the sensor array device is used together with the signal acquisition system to collect the wrist motion signals of subjects,and after data pre-processing,it is stored into the database.The establishment of the database lays the foundation for the next research on finger motion recognition based on artificial neural networks.3.The BP neural network model based on PVDF sensor array is constructed.This thesis builds a three-layer BP neural network model.And then taking the short-term energy value of the signals as the feature value and inputting them into the network for finger motion recognition,finally,getting the average recognition rate of finger movements which is 81.6%.The experiment indicates that the sensor array in cooperation with the BP neural network could effectively detect and recognize the semantics of finger movements.The above research shows that the PVDF sensor array designed in this thesis is small in size,easy and comfortable to wear.At the same time,it has a high recognition rate when applied to finger motion recognition.This device probably become an ideal device for finger motion recognition.