Path inference from sparse floating car data for urban networks

Abstract

The use of probe vehicles in traffic management is growing rapidly. The reason is that the required data collection infrastructure is increasingly in place in urban areas with a significant number of mobile sensors constantly moving and covering expansive areas of the road network. In many cases, the data is sparse in time and location and includes only geo-location and timestamp. Extracting paths taken by the vehicles from such sparse data is an important step towards travel time estimation and is referred to as the map-matching and path inference problem. This paper introduces a path inference method for low-frequency floating car data, assesses its performance, and compares it to recent methods using a set of ground truth data.