In the last decade we witnessed an increased demand for employment of unmanned aerial vehicles (UAV) in practise. For instance, there is a growing need to provide surveillance tasks in a given area by a team of cooperating UAVs. In this case, the ability of a single UAV to plan its course of actions (e.g., trajectories that the UAV must fly through) is essential. Trajectory planning algorithm used by UAVs must be able to find trajectories satisfying constraints given by environment (e.g., obstacles) or by UAVs’ dynamic models. Besides the planner itself the UAVs must somehow react to changes of high-level tasks or environment. Such a reaction often means to replan the trajectories towards new goals. In this paper, we will discuss the replanning related issues such as swapping the old and new trajectory smoothly respecting the UAV dynamics. We present an idea based on estimating running time of replanning tasks and evaluated its impact to safeness of replanning (e.g., avoiding to get to an inconsistent state).
展开▼
