Trajectory optimisation for obstacle avoidance in multiple fixed-wing UAVs formation flight using a modified GWO algorithm

Abstract This paper addresses the optimisation of the obstacle avoidance trajectory for unmanned aerial vehicles (UAVs) flying in a V-shaped formation. A control framework based on potential functions is utilised to maintain the formation and track the virtual leader. The obstacle avoidance mechanism is activated by defining an `activation circle’ and applying an optimal displacement vector to the virtual path. The grey wolf optimisation (GWO) algorithm is employed to determine the optimal paths, where the decision variables are the distances of the actual path points from the virtual path. The cost function is designed to simultaneously optimise three criteria: path length, dynamic turning radius and the number of obstacle collision points. To improve the algorithm’s efficiency, an exponential function is introduced for the exploration-exploitation balance control parameter in the GWO. Simulations demonstrate that this method is capable of generating optimal, short and collision-free paths while simultaneously respecting the dynamic constraints of the UAVs and maintaining the formation’s integrity. Potential operational applications, such as cooperative surveillance and search-and-rescue, are noted, and practical implementation aspects including robustness against wind disturbances and sensor noise are also briefly discussed.