Abstract
By discretizing the UAV's heading angle in a three-dimensional Dubins model, the path planning problem and the task assignment problem are established as a discrete graph model, aiming to solve the problem that decoupling solutions can only obtain local optimal solutions for multi-heterogeneous UAV task assignment and track planning. A genetic algorithm strategy based on parallel processing is established to realise the fast solution of the mixed integer programming problem; a depth-first algorithm is introduced to determine the feasibility of task planning by detecting the loop state of the timing constraint graph to eliminate the deadlock problem in execution. The simulation findings demonstrate that the integrated solution technique, taking into account the coupled task planning results, yields much more optimal planning results compared to the decoupling method's task planning outcomes. The distributed genetic algorithm has clear advantages over the traditional centralised genetic algorithm in terms of solution efficiency.