Optimal path planning for two UAVs in a pursuit-evasion game

Abstract

In this paper a path planning technique will be introduced for the unmanned aerial vehicles (UAV) fly at low altitude using a synthetic approach based on game theory and artificial neural networks. The low altitude pursuit-evasion maneuver of two UAVs - is defined based on an optimal control approach. Moreover, it has been sought to utilize optimal control rules and Differential Games theory to calculate the most favorable trajectories for both UAVs - one as the pursuer and the other as an evader. Since producing the optimal trajectories through solving the related equations may be a time-consuming trend, an artificial neural network is utilized to predict the flyable trajectories. The multilayer perceptron networks are trained using a set of trajectories obtained based on the differential game theory approach and could locate the position where the evader is captured. Hence, choices could made in real times. Consequently, the comparison of neural network results with accurate data obtained previously in the optimal control section confirms the accuracy and performance of the proposed method.