Multi-UAV Cooperative Search System Based On Information Interaction

Today,the industrialization process is accelerating,and the information environment is undergoing tremendous changes.In the past few decades,as a prominent representative of the third generation of intelligent robots,unmanned aerial vehicle(UAV)have taken the lead in the direction that from single campaign mode to multi-UAVs cooperative combat mode.Studies have shown that using multiple UAVs to search for an target area can expand the search horizon,improve the efficiency of the UAV to complete the mission,reduce the energy consumed in flight,reduce environmental instability,and further high-precision positioning to improve target hit rate.However,clustered UAVs are also facing some new problems.The robustness of multi-agent topology,the timeliness of online perception mechanism,the reliability of information interaction,and the loose coupling of collaborative systems are still challenging.The system provides a robust solution for the autonomous clustering and exploration mission of the UAV cluster to the target environment.In the implementation of the algorithm,a cluster search framework of UAV based on Anti-flocking algorithm is proposed to achieve efficient search of the target environment.Because of its fully distributed and intelligent control mechanism,the algorithm has better scalability,adaptability and robustness than stochastic motion model,and can adapt to dynamic environment more easily.In the flight environment,there are many environmental factors and interference factors.The computing power of the UAV hardware is weak.In order to better improve the sandbag delivery accuracy rate,this thesis combines support vector machine and k-means clustering algorithm to construct an online learning mechanism.In terms of building the platform,the DJI M100 quadrotor is used as the flight development platform,equipped with Manifold,ZenmuseX3 camera,STC15F2K60S2 MCU,S3003 steering gear and other modules.It meets the requirements of fast processing speed and more interactive interface.Using the portability and loose coupling of the ROS platform package,a multi-UAVs collaborative information interaction framework was built.The software system was divided into three nodes: visual detection and recognition,cluster control and flight task management,and detailed design details are given.A ground station software was designed for monitoring data to accept data processed by the flight platform and send control commands to control the order of execution of UAV missions.A set of communication protocols was provided for parsing the information interaction between the UAVs and ground station.Finally,in order to simulate the mission,the test was conducted in outdoors.The test results show that the whole system can work normally and complete the search and sandbag delivery task,and can better overcome the interference caused by weather,altitude and other factors.