This paper introduces a concept of safe path planning for UAV’s autonomous operation in an urban environment where GNSS-positioning may become unreliable or even unavailable. If the operation environment is a priori known and geo-localized, it is possible to predict a GNSS satellite constellation and hence to anticipate its signal occlusions at a given point and time. Motivated from this, our main idea is to utilize such sensor availability map in path planning task for ensuring UAV navigation safety. The proposed concept is implemented by a Partially Observable Markov Decision Process (POMDP) model. It incorporates a low-level navigation and guidance module for propagating the UAV state uncertainty in function of the probabilistic sensor availability. A new definition of cost function is introduced in this model such that the resulting optimal policy respects a user-defined safety requirement. A goal-oriented version of Monte-Carlo Tree Search algorithm, called POMCP-GO, is proposed for POMDP solving. The developed safe path planner is evaluated on two simple obstacle benchmark maps as well as on a real elevation map of San Diego downtown, along with GPS availability maps.

本文介绍了安全路径规划的概念用于无人机在城市环境中的自主运行，在该环境中GNSS定位可能变得不可靠甚至不可用。如果操作环境是先验已知且已地理定位，则可以预测GNSS卫星星座，从而可以预测给定点和时间的信号遮挡。因此，我们的主要思想是在路径规划任务中利用这种传感器可用性图来确保无人机导航的安全性。提出的概念是通过部分可观察的马尔可夫决策过程（POMDP）模型实现的。它包含一个低级导航和制导模块，用于传播概率传感器可用性函数中的无人机状态不确定性。在此模型中引入了成本函数的新定义，以使生成的最佳策略符合用户定义的安全要求。提出了一种面向目标的蒙特卡洛树搜索算法，称为POMCP-GO，用于POMDP求解。在两个简单的障碍物基准图以及圣地亚哥市区的真实高程图以及GPS可用性图上对开发的安全路径计划器进行了评估。