The Research On The Path Planning And Formation Tracking Technology Of UAV Swarm

In a complex battlefield environment,the proportion of unmanned warfare is gradually increasing.So,the demand for high-performance distributed collaborative technology for UAV swarms is increasingly urgent.This paper takes an approaching reconnaissance and target tracking mission as a research sample,aiming at three aspects: swarm migration trajectory planning before approaching,behavioral-level high-precision control during approaching,and coordinated reconnaissance and tracking of enemy targets after reaching the battlefield.The main contributions are as follows:(1)An online self-organizing path planning algorithm in complex terrain are proposed to solve the difficulties in swarm migration,such as obstacle and collision avoidance,population separation and aggregation,and emergency handling.It is based on the idea of ’reference track preplanning + global online adjustment’.First,the multi-objective particle swarm optimization algorithm is used to quickly pre-plan a flight path as the guidance of the swarm movement.Aiming at the uneven of Pareto front caused by terrain,a class of intra-domain vibration functions are designed to realize the mutual correction of collision fragments within the neighborhood of each particle and improve the convergence speed.Second,in the global online adjustment stage,the swarm is taken as a whole.Based on the individual cooperation method in bionics and swarm motion,the forward ellipsoid neighborhood design and dynamic neighbor selection were carried out to realize self-organization of migration track and improve the flexibility and real-time performance of task completion.Finally,the simulation results show that the proposed algorithm can guide the UAV swarm to the specified target safely and efficiently.(2)Aiming at the unfavorable factors in the flight process,such as communication delay and load variation,a high-precision adaptive sliding mode control algorithm is studied.It is based on,’strategy layer + behavior layer’,double formation control structure.The control algorithm ensures the formation accuracy of UAVs under variable load conditions and provides accurate load data for the strategy layer to improve decision accuracy.First,a delay state observer is designed for the communication delay existing in the internal communication of the swarm,whose current state is predicted based on the delay output of the neighbor.Second,a full-coefficient adaptive sliding mode controller is proposed,and the asymptotic convergence of the flight error of UAV swarms is proved based on the Lyapunov stability theory.Finally,simulation experiments show that the control protocol has excellent adaptability to the common variable mass and variable inertia behaviors and their adverse effects on system parameters in swarm flight missions.(3)Aiming at the situation of satellite’s signal loss of lock and insufficient positioning information in the cooperative reconnaissance and tracking tasks of non-cooperative targets,a cooperative control algorithm of UAV swarms based on bearing constraint theory is studied.First,the specific meaning of the bearing information in the field of view is clarified,and the UAV swarm tracking architecture is constructed based on the bearing graph theory.Then,a class of master-slave hierarchical UAV swarm control protocol is proposed based on the designed architecture.In the protocol,if the masters can realize target positioning and tracking based on vision,the slave can get rid of the dependence on the global position information,and realize positioning and movement based only on the master-slave bearing error.The protocols reduce the data transmission load in the cooperative reconnaissance,and ensures that UAV swarm still has the ability of formation keeping and changing in the absence of location information.Finally,Lyapunov stability theory and simulation experiments are used to prove the asymptotic convergence of swarm formation error.(4)Aiming at the situation of satellite’s signal loss of lock and insufficient real-time information acquisition capability in target cooperative reconnaissance and tracking tasks,a kind of master-slave layered sampling control protocol via bearing-based hybrid protocols is designed.In the protocols,event trigger and fixed time trigger are used to drive the UAV swarm respectively,so that each UAV can track the mission target at the same angular speed based on the sampled bearing information,and maintain the desired formation.This method can avoid the continuous monitoring of bearing error,and further reduce the pressure of acquiring bearing information and control frequency of swarm.Finally,the simulation results show that this control protocol can accomplish maneuvering target circumnavigation detection and tracking task well under the premise of effectively reducing communication load,and has strong practical significance.(5)Aiming at the low efficiency and high cost of swarm flight test,a high-fidelity visual experiment platform for UAV swarms is built,and the algorithms in each part are verified based on it.First,based on the visual experiment platform,the feasibility of online self-organizing flight path planning for swarm migration,adaptive sliding mode control for UAV formation task,swarm sampling control based on bearing constraint are verified respectively.Second,the feasibility of the visual saliency detection and the aerial target recognition based on deep learning are verified,which can be uesd to obtain bearing information.Finally,the proposed time-delay state observer and adaptive sliding mode controller,and the bearing constraint protocols of slaves are partially applied to the formation test in flight mission.