In this paper, we study a new data mining problem of obstacle detection from trajectory data. Intuitively, given two kinds of trajectories, i.e., reference and query trajectories, the obstacle is a region such that most query trajectories need to bypass this region, whereas the reference trajectories can go through as usual. We introduce a density-based definition for the obstacle based on a new normalized Dynamic Time Warping (nDTW) distance and the density functions tailored for the sub-trajectories to estimate the density variations. With this definition, we introduce a novel framework \textsf{DIOT} that utilizes the depth-first search method to detect implicit obstacles. We conduct extensive experiments over two real-life data sets. The experimental results show that \textsf{DIOT} can capture the nature of obstacles yet detect the implicit obstacles efficiently and effectively. Code is available at \url{https://github.com/1flei/obstacle}.
翻译:在本文中,我们从轨迹数据中研究新的障碍探测数据挖掘问题。 直觉上,考虑到两种轨迹,即参考和查询轨迹,障碍是一个区域,因此大多数查询轨迹都需要绕过这个区域,而参考轨迹可以照常运行。 我们根据新的正常的动态时间转换(nDTW)距离和为亚轨迹量测密度变化而定制的密度函数,为障碍设定了基于密度的定义。 根据这个定义,我们引入了一个新的框架\ textsf{DIOT},利用深度第一搜索方法探测隐含障碍。我们对两个真实数据组进行了广泛的实验。实验结果显示,\textsf{DIOT}可以捕捉障碍的性质,但能有效探测隐含的障碍。 代码可在\url{https://github.com/1fley/obstacle}查阅。