Flight Trajectory Generation For Multi-UAV Cooperation

The UAV's flight environment is more and more complex and the mission of flight is more and more high-demanding.The trend of development of UAV's technology is using information sharing and resource optimization to establish multi-platform autonomous mission coordination.In need to coordinated flight and trajectory generation of UAVs,based on theories in four-dimensional trajectory generation,this review puts forward an improved tau-H guidance strategy which is more suitable to meet the needs of the UAV trajectory generation and to solve the cooperative task.In order to improve the automation level of UAV and enhance the environmental adaptability of the track that it can be used for rapid decision-making through experience,this paper proposes a trajectory generation method based on reinforcement learning through the establishment of the state action with the improved tau-H guidance strategy.The feasibility of the method is verified by the simulation cases analysis.By introducing the correlative term of initial velocity into the motion formula,the improved Tau-H strategy overcomed the original shortcomings of common-used Tau-H motion strategy which was only applicable to problems with initial and ending velocity of zero.Therefore it can be employed to plan four dimensional(4D)motion of UAVs with nonzero initial and ending velocity to match the arrival time and velocity.Global trajectories are generated beforehand using particle swarm optimization(PSO)algorithm to search the initial coupling coefficient of UAVs.Based on the established communication topology among UAVs,the trajectories are constantly updated by solving the local planning problem by the rolling optimization method with double drive of sampling interval and collision detection.Simulations of several examples demonstrate the validity of the present method applicated to 4D coopertative trajectory generation and optimization of multi-UAV.In order to improve the flexibility and enhance the learning ability of trajectory generation,a method of four-dimensional trajectory generation using deterministic strategy gradient method is proposed.Through the improved Tau-H guidance strategy combined with reinforcement learning theory for state action mapping with trajectory,deep learning method and deterministic strategy gradient method are selected as the learning method of UAV 4D trajectory generation algorithm.The above research methods and results can provide some reference for multi-UAV coordination and trajectory generation.