An Improved GA-based Approach for UAV Swarm Formation Transformation

The unmanned aerial vehicles (UAV) swarm formation has drawn the attention of researchers in various military and civilian domains. Formation transformation is also a significant issue for the swarm of UAV. This article studies the transformation of UAV formation problem, and a modified formation transformation method based on genetic algorithm for UAV swarm in 3D environment is developed to improve the efficiency of UAV formation. In our method, firstly, variable mutation rate is used in the genetic algorithm to increase or reduce the mutation rate in the iteration of the algorithm. At the same time, the population is grouped before the crossover operation according to the fitness value. These strategies can not only maintain the speed of the algorithm, but can also avoid the premature convergence to the local optimum and find a better solution. Then, the shortest total flight length of UAV swarm and the shortest formation completion time are formalized as two different problems respectively. Comparison experiment indicated that our proposed methodology can provide a better solution in comparison to conventional methods and particle swarm optimization algorithm. Finally, the comprehensive experiment in UAV simulation platform, named XTDrone, is conducted and the feasibility of the proposed method is validated.