Multi-UAVs target assignment using opposition-based genetic algorithm with multiple mutation operators

The article presents a novel heterogeneous target assignment method for multiple curvature-constrained UAV. The heterogeneous targets involve two kinds of search direction constraints, namely, a predefined seasrch direction and free search direction. The number of UAVs’ kinematic constraints represented by Dubins vehicles are the same. In this work, maximization of the entire surveillance effectiveness is chosen as the objective of the target assignment problem, and the search benefit of each target is affected by the target value and the corresponding time of arrival. The problem could be formulated as a modified Dubins multiple travelling salesman problem with search constraints. To solve this challenging problem, the tailored genetic algorithm (GA) incorporated with the opposition-based learning technique and multiple mutation operators are proposed, denoted as OGA-MMO. By introducing the opposition-based learning technique into the evolutionary process, the global search capability is enhanced. Meanwhile, multiple mutation operators are developed to increase the probability of producing excellent genes. Besides, the double coding chromosomes are introduced to clearly describe the multi-UAVs target assignment problem, which can also make the decoding more easily. Finally, OGA-MMO is compared with regular GA on several multi-UAVs target assignment simulations. The comparison results show that the proposed method is more efficient and stronger in escaping from the local optimum in solving the multi-UAVs target assignment.