Autonomous Navigation And Landing Technology Of UAV Facing Low Speed Small Platform

In recent years,with the rapid development of visual network technology,the accuracy and speed of target recognition technology in autonomous driving,robotics and unmanned aerial vehicles have been rapidly improved.Future,unmanned aerial vehicle（UAV）will play more important role in more fields,and the visual part of the network system,as an essential part of unmanned aerial vehicle（UAV）,is widely used to conduct reconnaissance,automatic object detection and tracking target tasks,such as the corresponding visual network algorithm is particularly important,unmanned aerial vehicles（UAV）is the target detecting and locating the key to successful implementation.The application of target recognition technology to UAV can help UAV to quickly distinguish the specific terrain and make immediate decisions through the application of vision system algorithm in the rapid detection of targets.In addition,the application of target recognition technology is of strategic significance for realizing the autonomous arrival of UAV to the precise target location,completing flight tasks and ensuring flight safety.Therefore,the autonomous navigation and landing technology of UAV with target detection technology as the core is very important.The key technologies involved in this paper mainly include GPS data processing,serial communication,target identification and tracking and flight control.Firstly,GPS module is installed on low speed platform,and GPS signal is emitted through wireless module.Secondly,by receiving GPS signals,the UAV conducts rough positioning and flight control to make it reach the upper part of the mobile platform and make the mobile platform appear within the scanning range of the camera carried by the UAV.Then,in the Darknet＿ROS network,the camera module on the UAV identifies and tracks the target through YOLO algorithm,and predicts the moving trajectory of the target,constantly narrowing the distance with the target.Finally,control the UAV to land autonomously on the target platform.This paper uses DJI M210 v UAV to build embedded device NVIDIA Jetson TX2,wireless communication module ATK-LORA,GPS module ATK1218-BD and wide-angle 4K-7730 camera as the platform to verify this scheme.Experimental results show that the design of this paper realizes the functions of autonomous navigation,target recognition,tracking and autonomous landing when facing low speed small platform.The main research contents of this paper are as follows:（1）In order to reduce the relative error of GPS,the mobile platform is equipped with the same embedded device Jetson TX2,GPS module ATK1218-BD and wireless communication module ATK-LORA to obtain GPS data carried by UAV and mobile platform in real time.GPS data is transmitted through wireless module ATK-LORA.（2）In this paper,YOLOv4 target detection and recognition algorithm is used to obtain recognition results of high accuracy,fast recognition speed and real-time detection.（3）based on the above two points,according to the unmanned aerial vehicle（UAV）for mobile platform of GPS signal,deduces the coordinates of the mobile platform,the use of the flight control algorithm,control of unmanned aerial vehicle（UAV）make its constant decreases and the mobile platform distance,until the camera mobile platform in unmanned aerial vehicle（UAV）,when the camera to the mobile platform,using the target tracking algorithm to track mobile platform.In this paper,the uav and NVIDIA Jetson TX2 embedded device are developed to complete the system,and the Gazebo simulation experiment is used to verify the single visual distance algorithm,and the effect of the system is verified on DJI UAV M210 v platform.This paper has certain reference value to realize the real-time target recognition,autonomous navigation and landing of UAV.