Solving A Multi-objective Mission Planning Problem for UAV Swarms with An Improved NSGA-III Algorithm

Restricted communication in unmanned aerial vehicle (UAV) swarms means that configuration needs to vary dynamically with changing tasks. We propose a mission planning model that uses a motif, a grouping of related functions, as the basic task unit. The planning model automatically generates a mission planning scheme from a task priority execution order given as an input. The selection of the best scheme from among possible solutions is a multi-objective optimization problem with calculation complexity rapidly increasing with the number of tasks. To address this difficulty, we enhance the NSGA-III algorithm by adding adaptive genetic operators when generating the offspring population. We apply the improved NSGA-III algorithm to optimize mission planning schemes with changing task priority execution orders. We validated the feasibility and effectiveness of the improved algorithm via a case study.