UAV Cooperative Task Assignments for a SEAD Mission Using Genetic Algorithms
Marjorie Darrah, William Niland, Brian Stolarik, Lance Walp · AIAA Guidance, Navigation, and Control Conference and Exhibit · 2006
This paper discusses research that applies Genetic Algorithms (GAs) to efficiently allocate vehicles in a multiple task assignment problem during a Suppression of Enemy Air Defense (SEAD) mission. The assumptions for the mission include survivable vehicles (modeled after Unmanned Combat Aerial Vehicles (UCAVs)), heterogeneous teams, and dynamic target discovery. As targets are discovered throughout the mission the problem formulation must be dynamically updated to meet the new situation and the algorithm must be applied to find a new feasible and efficient solution. We compare the results of previous work done on the same mission scenario using Mixed Integer Linear Programming (MILP) to solve the assignment problem. While the MILP gives the optimal solution it requires much more time and computer resources to formulate and solve. Both approaches work very well for small scenarios with few vehicles and few targets. However, when the mission involves more than three vehicles and two targets, the GA can provide a feasible solution in a greatly reduced time. This paper discusses the GA approach, formulation, implementation, and integration into a high fidelity simulation.