Research On Patrol Monitoring Robot Based On Sound Feature Recognition Technology

With the improvement of living standards,people's awareness of safety and property protection has gradually increased,and the rapid development of social economy has made the scale of activities increasingly large.Traditional human patrols and fixed camera surveillance rely on human resources and limited monitoring.It is often the result of half the effort.This has been unable to meet the security requirements of today's society and cannot guarantee the security of the social environment.In view of the above problems,this project aims to design a low-cost,high-efficiency,all-round dynamic patrol monitoring robot.The work done on this topic is as follows:First,in order to realize the real-time positioning alarm function of the robot after monitoring the abnormal sound,a patrol monitoring robot system is designed.The system includes a monitoring robot and a host computer,and uses GPRS to perform long-distance autonomous wireless communication between the host computer and the robot,solving the problem that the WiFi signal is unstable and the Bluetooth communication distance is close;using the infrared pair tube to realize the robot path navigation,using RFID and GPS Indoor and outdoor positioning allows the robot to travel in a remote corner of the campus without losing it.The electret microphone and audio signal encoding chip are used to record the monitored sound,and the patrol position and sound monitoring data are packaged and sent to the host computer.Experiments show that after the abnormal sound is heard,the robot will respond and pop up a notification in real time on the host computer to remind the security personnel that the patrol monitoring robot can implement the patrol monitoring task.Second,in order to realize the real-time monitoring function of the robot,two sound monitoring scenarios are set.The first type only monitors whether there is sound,and presets the threshold.When the sound is monitored,the sound is recorded for 5 s,and it is floated and summed.When the threshold is greater than the threshold,the abnormal sound is displayed on the host computer and displayed.Robot position.Secondly,matching the monitored sound,the sound existing in the patrol environment is stored in the SD card in advance,as a normal sound,when the sound is monitored,the sound signal preprocessing,feature extraction and matching,matching are performed by using the MFCC-KNN method.The upper part is a normal sound,otherwise it is marked as an abnormal sound and displayed on the host computer and displays the robot position at this time.Experiments of two scenarios were carried out indoors and outdoors,and experiments showed that the sound monitoring can be realized in both cases,indicating that the patrol monitoring robot can perform patrol monitoring tasks.Third,in order to further improve the accuracy of the matching sound,the MFCC is improved,that is,threshold filtering is performed before the MFCC to improve the sound matching identification accuracy,reduce overlap,and eliminate noise.Compared with the traditional original MFCC,it is found that the method has a certain degree of improvement in sound matching,that is,the Euclidean distance value of the same sound is smaller,and the Euclidean distance value of different sounds is larger.At the same time,experiments on indoor and outdoor methods show that the method can realize sound monitoring and the sound matching effect is better.This paper designs a patrol monitoring robot system,realizes the autonomous mobile patrol function,and can perform sound processing on the robot,multi-sensor fusion and other technologies to form a comprehensive patrol system.Combined with the patrol and monitoring functions realized by it,and displaying the status in real time on the host computer,and using MFCC-KNN sound matching method and improved MFCC matching method to match the sound,it proves the practicability of the patrol monitoring robot and can complete the expected The basic patrol function,the accuracy meets the requirements of practical applications,and contributes greatly to the realization of the intelligent three-defense mode of "human defense + robot anti-fixed camera monitoring" in the future.It has certain reference value in the design of security robot products.