MAHACO: Multi-algorithm hybrid ant colony optimizer for 3D path planning of a group of UAVs

Abstract

Path planning is a critical part of unmanned aerial vehicle (UAV) achieving mission objectives, and the complexity of this problem is further increased when used for a group of UAVs. In addition, introducing curves based on different polynomials can design a smooth path for UAV that is continuous and meets safety constraints. Considering the above challenges, this paper proposes a multi-algorithm hybrid ant colony optimizer (ACO) named MAHACO, which is used for a 3D smooth path planning model of a group of UAVs based on the Said-Ball curve (SBC, for short). Firstly, by using the basic principles of other intelligent algorithms, ACO is extended to the continuous domain and three strategies are designed. Subsequently, the adaptive foraging strategy optimizes the ability of ACO to balance the exploration and exploitation phases and enhances its exploration ability in the search space. In addition, the multi-stage stochastic strategy expands the exploration range of ACO in the search space by enriching the selection of random vectors. Finally, the aggregation-mutation strategy improves the behavioral diversity and dynamics of ACO. To test the overall performance of MAHACO, it is compared with some state-of-the-art or improved metaheuristic algorithms on the highest dimensional CEC2020 and CEC2022 test sets, respectively. From the experimental results, the proposed MAHACO exhibits stronger performance advantages on 17 of the 22 functions. Then, the collision avoidance constraint and the communication constraint are introduced into the basic 3D path planning model of single UAV, and the model is extended to the application of a group of UAVs. This paper establishes a 3D smooth path planning model of a group of UAVs by taking the control points of SBC as the optimization variable of intelligent algorithms. Compared with other algorithms that rank high in the overall performance on the benchmark sets, MAHACO demonstrates its better practicability through basic and smooth path planning models, respectively.