Research On Forest Fire Monitoring And Path Planning Based On UAV

At present,UAV navigation and monitoring technology has been widely used in forest fire prevention and control fields,such as forest farm inspection,fire detection,flame fighting and so on.With the popularity of UAV in forest farms,real-time flame detection technology for patrol UAV and real-time path planning technology for fire-fighting UAV have become one of the most important and challenging topics in UAV forest fire prevention system.For the patrol UAV,because the forest image information transmitted by the UAV is complex,the missed detection rate of manual observation is high.Therefore,to make the UAV identify the flame quickly and accurately in the complex environment is the key point and diffioulty to roalize the roal-time autonomous monitoring of the patrol UAV;for the fire-fighting UAV,because of the high flight speed of the UAV,it is difficult to operate the UAV manually;the forest environment is complex,so it is difficult to obtain complete map information.Therefore,in the case of incomplete environmental information,the UAV can quickly plan a collision-free optimal path from the starting position to the flame position,which is the key step to realize the autonomous fire fighting of the fire-fighting UAV.For this reason,this paper takes the Tello UAV as the research object,and deeply studies the real-time fire detection,and fire-fighting UAV real-time path planning in the forest firo prevention and control system.The main work is as follows.(1)The scheme of real-time flame monitoring and fire extinguishing system of UAV for forest fire is constructed.The forest fire monitoring and fire extinguishing system is composed of monitoring UAV,fire extinguishing UAV and ground console.The forest fire monitoring and fire extinguishing system consists of two parts:the first part is used to monitor forest fire,and the monitoring UAV and ground console are used to complete the task of flame detection and fire location;the second part is used for fire fighting UAV path planning.according to the first part of the detection results,quickly plan the optimal collision-free path from the starting position to the fire point.(2)Aiming at the problem of forest fire monitoring,a UAV flame monitoring platform is constructed.Under the theoretical framework of YOLOv4,a real-time flame detection algorithm for UAV is proposed,including flame target detection and fire location.First of all,the ground console generates the monitoring trajectory according to the ground shape and the parameters of the UAV.Subsequently,the monitoring UAV flies according to the trajectory and sends the image data obtained by the camera to the ground console.The ground console uses the proposed algorithm to monitor whether there is a flame in the image.If the flame is detected and the inspection mission is interrupted,the ground console calculates the position of the flame.Finally,in the outdoor scene,several groups of UAV real-time flame detection experiments are completed by using Tello UAV.The experimental results show that the UAV flame detection algorithm based on YOLOv4 can accurately identify the flame and mark its position under different flame color saturation conditions,such as sunny day,cloudy day,rainy day and so on.(3)Aiming at the problem of fire-fighting UAV path planning,a fire-fighting UAV control platform is constructed,and a new improved D*Lite algorithm is proposed,which combines HPA*,on the basis of D*Lite to greatly improve the speed of dynamic path planning.At the same time,two smoothing algorithms are added to reduce the number of turning angles and the amplitude of turning angles of UAV in flight.Subsequently,this paper carries out several groups of UAV path planning experiments in the simulation environment,and the experimental results show that the algorithm can quickly and effectively plan the safe path for the UAV to reach multiple static or dynamic targets.Finally,under the indoor natural scene,the path planning experiment of UAV is completed by using Tello UAV.The experimental results show that the improved D*Lite algorithm can meet the real-time requirements of fire-fighting UAV path planning and the accuracy requirements of UAV reaching the flame point quickly.