Intelligent Patrol Inspection UAV System For Urban Security Scenarios

In recent years,unmanned aerial vehicles(UAV)have been widely used in Plant Protection,Electricity Inspection,Highway Patrol and other industries for their advantages of high flexibility,strong real-time performance and large monitoring range.With the continuous expansion of its application scenarios,this industry is marching for a promising and broad prospect.City security is also one of the important areas in the application of drones.UAV greatly ease the pressure of public security management in the security of large-scale events such as Spring Festival travel and evening parties,or daily inspections such as epidemic prevention and control.However,UAV only plays the role of video collection in industrial applications at this stage and it still needs a lot of manpower and time to complete the image content judgement.With the rapid development of artificial intelligence,computer vision algorithms can deeply understand the scene and provide further decision support.In this context,it`s urgent for UAV technology to integrate with AI technology deeply to expand the job of UAV reconnaissance and give itself a certain autonomous response capability.Forging UAV system with the existing mature security inspection businesscan realize long-distance perception analysis,warning instance prediction and autonomous decision-making.Main research contents of this paper are as follows based on the envision above:1.Design a set of automatic UAV flight based on terminal computing equipment after analyzing the requirements of the UAV inspection system and investigating the existing UAV inspection application solutions in the industry in combination with the real situation of the security scene.2.Research and design a set of hardware platform system in which UAV plays the role of main body,including UAV flight controller,on-board computer and image transmission equipment.Dig into hardware performance parameters and related technologies.Realize the goal of UAV’s fixed-point and fixed-distance flight under the control of codes.3.To solve the problem of slow calculation speed of the original AI algorithm on mobile terminal devices,improve the network structure of the action recognition algorithm by analyzing the cutting-edge target detection,motion detection and target tracking algorithms so the algorithm can run smoothly on the on-board computer carried by the UAV and detect the real-time picture.Besides,achieve the goal of tracking the real-time target and flying with it after combing the tracking algorithm with the UAV control instructions.We design a set of UAV intelligent inspection system in security scenarios,which can identify suspicious targets in urban streets and actively track them.The finished system can run stably on multiple sets of UAV equipment with different configurations.The experimental results show that:It can realize remote control with a radius of 1000 meters after establishing a direct link between the UAV terminal and the ground control host;The motion detection algorithm can run on the terminal device at a speed of 10 fps after frame sampling training;The accuracy of the detection algorithm can reach to 90% after integration;The UAV can automatically track objects according to the detection results of the visual algorithm.The research results of this paper provide a reference scheme for deploying visual algorithms on UAV equipment.