DIOT: Detecting Implicit Obstacles from Trajectories
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}.