Unmanned Aircraft Systems (UAS) provide the platform of many applications such as efficient package delivery, monitoring, surveillance, search and rescue, and more. As the density of UAS deployed in urban areas for delivery and monitoring operations is expected to increase in the near future, it becomes critical to design an efficient collision-free UAS path planning system to be used by a centralized traffic manager. In this article we propose an on-demand, scalable, collision-free trajectory planning mechanism that processes routing requests in batches, with the objective of improving the ratio of successfully admitted requests. The planner transforms the input map to a unit graph discretized on the safety distance and computes trajectories using search in the derived space-time graph. The planner generates trajectories that are by design conflict-free and deadlock-free. The performance of our algorithm is analyzed with simulations. We measure admission ratio, the delay overhead vs. optimal path delay, the algorithm running time, and we compare results with prior work. The proposed batch planner yields better admission ratio and delay overhead in low and medium density scenarios. In all scenarios it has a running time that is 4-7 times shorter, showing its scalability in practical situations.

中文翻译：

---

利用时空图进行 UAS 批量路径规划


无人机系统 (UAS) 提供高效包裹递送、监控、监视、搜索和救援等许多应用的平台。由于在城市地区部署的用于交付和监控操作的无人机密度预计在不久的将来会增加，因此设计一个供集中式交通管理器使用的高效无碰撞无人机路径规划系统变得至关重要。在本文中，我们提出了一种按需、可扩展、无碰撞的轨迹规划机制，可以批量处理路由请求，目的是提高成功接纳请求的比率。规划器将输入图转换为在安全距离上离散化的单位图，并使用派生时空图中的搜索来计算轨迹。规划器生成的轨迹在设计上是无冲突和无死锁的。我们的算法的性能通过模拟进行了分析。我们测量准入率、延迟开销与最佳路径延迟、算法运行时间，并将结果与​​之前的工作进行比较。所提出的批量规划器在中低密度场景中产生更好的准入率和延迟开销。在所有场景下，它的运行时间都缩短了4-7倍，显示了其在实际情况下的可扩展性。