Vision-based Cooperative Navigation And Dynamic Platoon Control For Multiple Unmanned Aerial Vehicles

In recent years,multiple unmanned aerial vehicle(multi-UAV)systems,which can complete complex flight tasks and improve overall work efficiency,have been a hot research area due to their outstanding advantages such as wide detection range,high payload capacity and strong environmental adaptability.Autonomous cruise is one of the basic ways for multi-UAV systems to perform cooperative flight tasks,which plays an important role in military exercises,traffic control,regional reconnaissance,and power inspections.During the multi-UAV cooperative cruising,the navigation system should provide reliable localization information and the target cruise path.The navigation method based on onboard vision sensors has the advantages of low cost,low power consumption and low mass,which is suitable for small-sized UAVs.In the actual cruising process,it is not only necessary to consider the tracking and control of the cruise path,but also to achieve mutual coordination within the UAV platoon while coping with the issues caused by the complex surrounding environment and emergencies for the task execution.This thesis focuses on the visual navigation and cooperative control of multi-UAV cooperative cruise systems.The main contributions of this thesis can be listed as follows:(1)A visual-inertial autonomous localization system for UAVs is designed to provide reliable positioning estimation for cruise missions,which can work in the GPS-denied environment.Considering the inter-vehicle status awareness requirements in the UAV platoon,the relative positioning information of multi-UAVs is obtained through the back-end optimization,so that each UAV can obtain its position in the global coordinate system.A multi-UAV collaborative positioning experiment platform is built to run the proposed algorithm in real-time,and the results are compared with the true value recorded by the motion capture system,which verifies the accuracy of the cooperative positioning algorithm designed in the thesis.(2)In order to achieve cooperative cruise tasks,a path planning strategy based on airborne visual sensors is proposed,in which the map can be built in real time.Then,an improved A*algorithm is developed to meet the lack of smoothness and dynamic constraints of the UAV through dynamic search and discretization of the control space,and the improved cost and heuristic function make path evaluation more accurate.A map is constructed in the actual environment and the path planning algorithm is tested,and the effectiveness of the proposed path planning algorithm is demonstrated through simulation results in the complex virtual environment.(3)The cooperative cruising control and dynamic obstacle avoidance problems for unmanned aerial vehicle platoon are investigated.The Transverse Feedback Linearization(TFL)method is employed for the platoon to follow the desired path.Following the cruise path,a distributed control strategy based on the consensus protocol is developed.In order to ensure the safety of the UAV platoon which is subject to the moving obstacle,this thesis also proposes an obstacle avoidance control strategy via the combination of the potential field methodology with the consensus protocol.The stability of the proposed control strategy is proved via Lyapunov based stability analysis and LaSalle’s invariance principle.Meanwhile,it is proved through the energy-based analysis that no collisions occur between any agent in the vehicle platoon and the moving obstacle.Finally,real-time experiments of cooperative flight control and obstacle avoidance are performed to demonstrate the effectiveness of the proposed control strategy.