Path planner methods for UAVs in real environment

Abstract This paper presents two methods to plan the path of an unmanned aerial vehicle (UAV) in a real 3D environment. Each method has its own purpose: one method is off-line and must allow an external operator to choose a trajectory to fly before the takeoff of the UAV; the other is on-line and must allow an external operator to modify the UAV trajectory in real-time during the fly. This last method is based on the well-known A-Star algorithm whereas the first one is a genetic algorithm (GA). The method based on A-Star provides a single path and the GA provides multiples trajectories using a Pareto front (PF). The paths produced have to satisfy the dynamic properties of the vehicle. Our methods embed the 2D dynamic properties of the vehicle and computes 2D trajectories. The third missing dimension is then recomputed using a recursive algorithm. The characteristics of the optimal path are represented in the form of multi-objectives. We achieve real-time capability for our A-Star based method with an average computation time of 836 ms and we achieve performance and flexibility for our GA.