Receding Horizon Control for UAV Formation Reconfiguration: An Enhanced Marine Predators Strategy

The uncrewed aerial vehicle (UAV) formation reconfiguration problem, due to its various limitations and high nonlinearity, has become a significant challenge in UAV formation. The existing methods mainly convert the problem into a constrained global optimization problem using control parameterization and time discretization (CPTD). This method, however, overlooks possible constraints on global optimization and the fact that solving the optimization problem itself can be difficult and time-consuming. In this paper, we firstly comprehensively considered the constraints in the UAV formation reconfiguration and proposed a problem model closer to real-world conditions. Then, we propose a new heuristic algorithm, the enhanced marine predators algorithm (EMPA) to address the constrained optimization problem, which has stronger optimization capability and greater stability when solving high-dimensional nonlinear problems. Finally, to decrease the computation cost, a more accurate and real-time receding horizon controller is designed based on the proposed EMPA, where the global optimal control problem of UAV formation reconfiguration is decomposed into a series of online local optimization problems. Simulation experiments verify the feasibility and effectiveness of the proposed control method. Note to Practitioners—This paper was motivated by the formation reconfiguration problem of uncrewed aerial vehicles (UAV). Existing approaches convert the problem into a constrained global optimization one using control parameterization and time discretization, which not only overlooks possible constraints on global optimization, but also has great difficulty in finding a feasible solution. This paper comprehensively takes into account such optimization constraints to build a problem model closer to real-world conditions, resulting in an enhanced marine predators algorithm of stronger optimization capability, greater stability as well as lower complexity. In this paper, a new heuristic algorithm called is developed, based on which a new receding horizon controller is established with its effectiveness validated through comparative experiments with other well-performing heuristic-algorithm-based controllers on the CEC-2017 benchmark. In future research, we will address the formation reconfiguration problem with leader-follower communication constraints, e.g. cyber-attacks, time-delays and topology failures.