Intelligent foresight for UAV routing problems

This paper describes an Intelligent Foresight (IF) method for improving the response time of a UAV routing optimisation, using the optimal solution of a smaller size scenario, as a warm-start for the Ant Colony Optimisation. The method aims to predict the revised routing and allocation of UAVs in the event of a new task arising. The new task location is not known: solutions are calculated for a range of possible locations. Therefore, a smart way of sampling and tessellating the plane is introduced to distinguish the areas where the solution will not change if that task is added. This hybrid optimisation was created to offer near-optimal solutions, under the condition that its computational time would be less than an exact method would require to solve only one scenario. The IF method was tested for a variety of scenarios and benchmarked against the Gurobi software. The results showed that the IF offers a good approximation of how the solution will change with its computational time remaining approximately the same, regardless the size or the complexity of the scenarios solved.