Article Preview
TopIntroduction
In recent years, with the development of UAV technology and its wide application in the military field, people have aroused new interest in the research of 3D trajectory planning of UAV. The problem of UAV flight path planning is to find an optimal route according to the requirements of specific flight plan and working environment. It is an important factor to ensure the safety and stability of UAV operation and realize unmanned flight.
UAV penetration is based on the effective path planning method, and its core depends on the merits of the optimization algorithm. For the UAV path planning problem in low altitude penetration mode, in recent years, domestic and foreign researchers have made some achievements. Research methods can be roughly divided into traditional methods and intelligent methods. Traditional methods such as fast-expanding random trees (Zucker, 2007), A* algorithm (Yang & Zhao,2004), etc. were initially widely used in 2D path planning. However, 3D paths are more effective in the field of UAV low-altitude penetrating ability and terrain tracking (Martin, 2016). At the same time, intelligent methods have made significant progress in the task of UAV trajectory planning. For example, literature (Shorakaei & Gholami, 2016) combines parallel genetic algorithm and probability map to complete the trajectory planning of coordination among multiple UAVs. This method overcomes the prematurity of genetic algorithm, but the performance constraints of UAV itself are not considered. Ma (Ma & Zhou, 2015) used genetic algorithm to determine the location of threat points and flight path points through polar coordinates, and reduced the dimension of the aircraft code, which simplified the search space and improved the planning efficiency. However, due to the inability to predict the length of the track, the coding length of chromosomes is also uncertain, and the dynamic adjustment is more complicated. The literatures (Ma & Zhou, 2005), (Fu et al., 2011) all use the particle swarm optimization method to solve the 3D trajectory planning problem. And the search space can be reduced by incorporating constraints into the pre-established cost function to evaluate whether particles are good or not. However, when faced with complex problems, the algorithm will converge prematurely and obtain a local optimum. (Ren et al., 2001) used the neural network algorithm to model the potential field through serial simulation in solving the trajectory planning problem. With the iteration of the neural network, the distribution of the potential field was updated by changing the connection weight coefficient, showing strong environmental adaptability. But this approach does not take into account other threats from the external environment. Reference (Tao & Wang, 2016) uses the chaotic particle swarm optimization algorithm, and Logistic chaotic map is used to make the algorithm jump out of the local optimum, which is applied to the trajectory planning of the parafoil system. (Liang & WANG, 2021) used the improved ant colony algorithm combined with the grid method to plan the fire evacuation route, but the application scenario of this problem is limited to the two-dimensional environment. In addition, (Storn & Price, 1997) (Zhou & Luo, 2013) use the difference algorithm, (Wen, 2005) (Wu et al., 2018) uses the ant colony algorithm, (Xu & Duan, 2010) (Wu & Li, 2021) use the artificial bee colony algorithm, etc. The wolf pack algorithm was first applied in literature (Liu, 2015) to solve the 3D trajectory planning problem of UAV, but there is still room for improvement in algorithm accuracy and convergence speed. Compared with traditional methods, swarm intelligence algorithms have more powerful global search capabilities and are more effective in dealing with complex real-world problems. However, in practical problems, many constraints need to be considered, and the objective function will be more complex, so it is particularly important to find a suitable and high-quality solution.