Real-time Mission Planning For UAVs Using Multi-phase Path Prediction

Unmanned Aerial Vehicle also known as UAV, because of its advantages of low cost, strong adaptability, good concealment and more and more widely used. UAV’s biggest advantage is that its cost of research, development, manufacture and application is much lower than other kind of aircraft. Therefore it has broad prospects in military and civilian application. In the case of a single intervention, to achieve a complicated task is the goal of UAV system in the future. As the single-handedly ability of the UAV to complete tasks is more and more strong, UAV collaborative ability should also be stronger and stronger, more study of UAV mission planning is very necessary.Mission planning problem is the key of UAV autonomous control, and the result of mission planning depends on the accurate prediction. However, in real scenario, all kinds of uncertain factors will bring large deviation. The group of UAVs mission planning must be based on task assignment subsystem, trajectory generation subsystem and navigation control subsystem is composed of three level structure to complete, and relations between levels to a large extent determines the results of mission planning. This paper aims to improve the planning efficiency by improving the mission planning system.The multi-phase path prediction method includes the following stages: the path estimation, the line path planning, the smooth path generation and the generation of the rendezvous trajectory. In each planning cycle, the results of each phase will be fed back to the task assignment subsystem, and then optimize the mission planning results. Path estimation stage is based on local A* algorithm to compute the paths between all UAV and all targets. In phase II, the results is obtained basing on the global A* algorithm. Trajectory smoothing stage is based on the B-spline method. The planar motion and the height variation are decoupled into account. In some scenario, many-to-one tasks are included and the trajectories need to be modified. The modified process is based on the Dubins model, which has the characteristics of small calculation, and can be estimated accurately.Overall evaluation of mission planning includes performance and efficiency. As a kind of real-time method, the former decides if the algorithm can get good result and the latter decides if the algorithm can be used for real-time application. In this paper, the simulation research includes simulation for performance research, simulation for efficiency research and simulation for scenario with many-to-one task.