Accurate Detection of Road Network Anomaly by Understanding Crowd's Driving Strategies from Human Mobility

摘要

There are thousands of road closures and changed traffic rules that impact vehicle routing every day. Detecting the road closures and traffic rule changes is essential for dynamic route planning and navigation serving. In this article, we propose a driving-behavior modeling-based method for accurately and effectively detecting the road anomalies. In the first step, we detect the areas of anomalies by using the deviation between drivers' actual and expected behaviors. To discover the cause of anomalies, we explore the drivers' short-term destination and find the crucial link pairs in anomalous areas through a novel optimized link entanglement search algorithm, namely, the Select Link Entanglements (SELES) algorithm. Finally, we analyze the crowd's driving patterns to explain the road network anomalies further. Experiments on a very large GPS dataset demonstrate that the proposed approach outperforms the existing methods in terms of both accuracy and effectiveness.